AI Investment on Uncle Xiang's Notebook

In a low interest rate environment, I have rearranged a ten-year financial plan.

Wed, 22 Jul 2026 20:00:00 +0800

After working for ten years, I once allocated time deposits during a period of relatively high interest rates, and later bonds became a more important part of my portfolio. The time deposits and bond products I held could provide a return of about 4.5% at one point. At that time, I was accustomed to putting my money in an appropriate term and letting time do most of the work.

Recently, the returns on the cash management products in my account have noticeably declined, and my past approach needs to be rearranged. Over the past six months, I have been reducing my pure bond positions and increasing my allocation to “fixed income+” products. Here, the “+” is not intended to make the return figures look more attractive, but because a single low-volatility asset may not cover the target for the next period of time.

What I want to organize is how to arrange a ten-year fund for myself when conditions such as low interest rates, the inability to guarantee the subscription status of cross-border funds, and the possibility of a pullback in the equity market all exist simultaneously.

From yield changes back to the cash flow question

Past interest rate experience easily leads me to think of asset allocation as “picking the place with a higher return.” When the low-risk return in my account drops, I start by asking myself: over the next five to ten years, does every sum of money I am certain I will need have its own place; can the money I won’t use for a long time tolerate net value fluctuations; and when fluctuations occur, do I still have a cushion left in the cash I hold.

I set my investment horizon to start at five years, and treat a decade as a planning cycle. This is the timeframe I give myself: it allows different market phases to all have a chance to appear on the books, and it accounts for volatility and recovery periods in my expectations. The horizon itself does not replace risk judgment, nor does it commit to recovering returns by some specific date.

For me, the sequence of fund planning takes precedence over the specific investment choices: first set aside funds for near-term expenses and emergency cash, then establish acceptable position limits for different categories, and only afterward discuss long-term systematic investments. When encountering a major market downturn, whether one can continue to stick to the plan often depends on whether there is still cash on hand that does not need to be liquidated.

I previously wrote an article on funds and fixed-income investments, documenting my experiences with funds and fixed-income products at the time; this old record also serves as a reminder that capital allocation should be re-examined based on my own stage in life and market conditions.

Real-world Constraints of Two Cross-Border Configurations

In the long-term equity portion, I have allocated fund shares that track the corresponding indices of the Nasdaq 100 and S&P 500, and participated through small regular investments. When arranging this portion of funds, I encountered situations where the subscription status of QDII funds could not be guaranteed. Therefore, “buy the dip” is not an action that can be executed at any time in my plan.

When the market is at a high level, I am reluctant to invest too much at once for fear of missing out; when a clear pullback actually occurs, the funds previously set aside may not be able to enter due to the subscription status. Market fluctuations and such subscription statuses are not within my control, so I cannot write them into a position-adding plan that is guaranteed to be executable.

Therefore, I only regard the cross-border portion as a limited seat in my long-term plan, pre-constraining the total allocation and the rhythm of regular investments. This arrangement cannot change market valuations or subscription availability, but it prevents treating funds that must remain liquid as though they were freely available for averaging down at any time.

Regarding the CSI 300 portion, I currently hold units of the corresponding index fund, and plan to gradually adjust to a CSI 300 enhanced strategy going forward. The arrangement of my base currency funds can be structured around my own cash flow, systematic investment pace, and reserved funds; if a drawdown occurs in the future, whether to increase investment will still be subject to the position cap determined in advance. This is the logic of my own plan and does not constitute a trading instruction for anyone.

Reviewing History with a Set of Reproducible Standards

My actual dollar-cost averaging (DCA) uses the platform’s smart DCA feature, which buys more in bear markets and less in bull markets. However, I don’t have access to Alipay’s algorithm parameters, nor do I have records of each actual deduction, so the backtest below does not replicate the Alipay smart DCA algorithm, nor does it represent actual account returns.

To make the results verifiable, the backtest uses a more朴素 alternative rule: on the first available net value date of each calendar month, invest an equal amount of 1 cash unit, converting to shares at that day’s unit net value, and value at the end of the period using the last available unit net value.

Annualized Return: Calculated as annual XIRR based on all monthly invested cash flows and ending account value.
Maximum Drawdown: Calculated based on the unit net value series over the fully available interval for the fund.
Drawdown Recovery Time: The calendar days from the net value high point before the maximum drawdown, until the unit net value first recovers to or exceeds that high point.

The NAV comes from the public historical NAV interface of East Money Tiantian Fund, fetched on July 22, 2026, using unit NAV rather than accumulated NAV or daily growth rate. Each fund uses its own full available range, and no long-term data has been truncated to make the comparison look complete.

This article uses unit net value as the data source calibration, without separately simulating account-level fees and frictions, including subscription and redemption fees, taxes, currency conversion costs, actual subscription restrictions, tracking errors, and other trading frictions; this also does not change the boundary that “equal monthly investment is merely a transparent alternative rule, not a reproduction of Alipay’s smart regular investment.”

Fund Code (Fund Name)	Backtest Period	Monthly Investment Count	Annualized XIRR	Maximum Drawdown	Drawdown Recovery Time
001015 (Huaxia CSI 300 Index Enhancement A)	2015-02-10 to 2026-07-21	138	7.61%	-42.02%	1,728 days
000051 (Huaxia CSI 300 ETF Link A)	2009-07-10 to 2026-07-21	205	5.77%	-43.43%	1,857 days
270042 (GF NASDAQ-100 ETF Connect RMB (QDII) A)	2012-08-15 to 2026-07-20	168	17.11%	-31.18%	588 days
019305 (JPMorgan S&P 500 Index (QDII) RMB C) *	2023-09-01 to 2026-07-20	35	13.95%	-17.44%	127 days

019305 has only 35 natural months of samples, having experienced significantly fewer market phases than the other three funds; its annualized XIRR, maximum drawdown, and recovery time only describe this short sample period. It cannot be compared horizontally with long-term samples, and even less should it be used to infer that future drawdowns will be smaller or recovery will be faster.

I first look at the maximum drawdown and the time it takes to recover. Taking this set of historical samples as an example, 001015’s maximum drawdown started from the peak on June 12, 2015, and it took 1,728 days for the unit net value to reach or exceed that peak again for the first time on March 5, 2020; for 000051, it took 1,857 days. Facing unrepaired unrealized fluctuations for several consecutive years is a scenario that this ten-year plan must accommodate.

Backtesting Didn’t Make the Decision for Me, But It Changed How I Frame the Question

Backtesting transforms my originally vague expectations into several questions that need to be answered in advance.

First, whether the reserved capital is sufficient to cover a repair period longer than expected. Drawdowns are not just weeks-long dips on a line chart—at least in the two CSI 300 samples mentioned above, it took more than four years to recover the previous high. If that portion of money is preset for short-term use, it becomes difficult to stick with the plan.

Second, whether the top-up assumption for cross-border assets depends on a subscription status that cannot be guaranteed. I keep existing positions, their possible fluctuations, and the fund arrangements of other assets within a range I can bear; top-up is merely an option when subscription conditions permit.

Third, whether the drawdown rules for my own principal have clearly defined boundaries. I will first write down the total amount of this portion of funds, how many installments to invest, and to what extent the price drops before I stop adding more; until this is clearly written down, I won’t treat temporary position increases as a rule.

In my experience with low interest rate environments, no single asset can return the past feeling of returns intact. My adjustment is to put bonds back into the stabilizing portion of the portfolio, place equities and cross-border allocations into longer cycles, and incorporate subscription status, cash flow, and drawdowns into the plan altogether.

A five-year starting commitment and a ten-year cycle is the time I’m willing to reserve for this plan. It is not a profit guarantee letter. When the next market fluctuation comes, I hope I won’t have to make a sudden decision about whether to sell off emergency funds, nor will I have to rewrite the entire plan due to changes in the subscription status of QDII funds I encounter.

Data Description

This article conducts backtesting using publicly available historical net value data from East Money Tiantian Fund: 001015 (Huaxia CSI 300 Index Enhanced A), 000051 (Huaxia CSI 300 ETF Linked A), 270042 (GF NASDAQ 100 ETF Linked RMB (QDII) A), 019305 (JPMorgan S&P 500 Index (QDII) RMB C). The net value was fetched on July 22, 2026, and the available data ranges and calculation methods are listed in the table below.

Backtests are calculated solely based on the unit net value from the data source, without separately simulating account-level expenses and frictions, including subscription and redemption fees, taxes, currency conversion costs, actual subscription limits, tracking error, and other trading frictions. Historical data does not represent future performance; the personal capital arrangements mentioned in the article are for illustrative purposes only and do not constitute investment, trading, or allocation advice.

References

Eastmoney Daily Fund Historical NAV API: https://api.fund.eastmoney.com/f10/lsjz (scraped on 2026-07-22; based on unit NAV)
Eastmoney Fund Pages: https://fund.eastmoney.com/001015.html, https://fund.eastmoney.com/000051.html, https://fund.eastmoney.com/270042.html, https://fund.eastmoney.com/019305.html (verified fund names on 2026-07-22)

写作附记

Original Prompt

$blog-writer Thinking about financial allocation in the era of interest rate cuts. Having worked for ten years, most of the time I experienced the high interest rate era. In the first four years, relying on fixed-term deposits could yield decent returns, and in the following five years, relying on bonds was also okay; 4.5% was not difficult in the past. With changes in the international environment, China has accelerated into an interest rate cut cycle, and the annualized return of Yu’ebao has fallen below 1%. In the past six months, I have been adjusting my bond position by reducing it and increasing my allocation to fixed income+ bonds. I configured Nasdaq 100 ETF and S&P 500 ETF for regular fixed investment. QDII foreign exchange quotas are restricted, so I made small regular investments. They are currently at high levels and not suitable for large amounts. There is a problem with US stock index ETFs: when there is a sharp drop, the quota is used up, so how to add to the position. I also allocated the CSI 300 Index and subsequently adjusted it to the CSI 300 Enhanced category, which is easier to control. When encountering pullbacks, add to the position, and capital planning needs to be reasonable. Follow the plan and run a backtest on returns. The regular fixed investments are all intelligent fixed investments, buying more in bear markets and less in bull markets, using Alipay’s algorithm. Organize the backtest data into a table and place it in the main text: annualized return, maximum drawdown, and drawdown recovery time. The fund codes used for the backtest: CSI 300 ETF Enhanced: 001015, CSI 300 ETF: 000051, Nasdaq 100 ETF: 270042, S&P 500 ETF: 019305.

The new investment plan has an investment cycle starting from five years, with ten years as one cycle.

Deregulation and Funding News Appear on the Same Day — How Zhipu and MiniMax Are Being Repriced

Fri, 10 Jul 2026 21:20:00 +0800

At the close on July 10, Zhipu (02513.HK) fell 19.29%, while MiniMax-W (00100.HK) dropped 9.68%. These numbers easily lead to a simple conclusion: the financing news was useless. However, the intraday trading did not follow a straight line. From morning to early afternoon, MiniMax-W’s decline was once deeper; in the afternoon it rebounded, while Zhipu continued to slide lower, ending the day with a larger drop.

In the same week, the two companies also faced newly unlocked shares and new financing arrangements. There are three different prices here: the marginal transaction price in the secondary market on that day, the price for placing new shares with specific investors, and the price at which convertible bonds may be converted into shares in the future. Mixing them into a single “market valuation” will distort the conclusion.

First, Look at the Price Path

The lock-up releases did not all occur on the same day. According to media statistics from Hong Kong Stock Decoder (港股解码), approximately 25.6816 million cornerstone investor shares of Zhipu were released on July 8, accounting for about 5.76% of the total share capital; approximately 146 million shares of MiniMax were released on July 9, accounting for about 46.44% of the total share capital, with its freely tradable shares prior to the release being less than 6%. This is a media statistic, not a company disclosure of actual reductions. It illustrates the magnitude differences in potential supply, but does not prove that these shares were sold on that day.

The daily chart first provides an overview. On July 9, the first trading day after the lock-up expiry, Zhipu closed higher while MiniMax-W closed lower; by July 10, both stocks declined, with Zhipu showing a larger trading volume and a bigger drop at close.

Date	Company	Open	High	Low	Close	Change	Turnover
Jul 8	Zhipu	1,563.0	1,918.0	1,450.0	1,825.0	+13.35%	N/A in this article
Jul 8	MiniMax-W	323.8	389.8	311.0	362.6	+12.0%	N/A in this article
Jul 9	Zhipu	1,879.0	2,222.0	1,803.0	2,032.0	+11.34%	N/A in this article
Jul 9	MiniMax-W	359.8	397.4	283.8	297.4	-17.98%	N/A in this article
Jul 10	Zhipu	1,850.0	1,999.0	1,597.0	1,640.0	-19.29%	Approx. HKD 13.260 billion
Jul 10	MiniMax-W	280.4	291.6	248.4	268.6	-9.68%	Approx. HKD 3.594 billion

The price unit is in HKD; the change percentage is calculated based on the previous trading day’s closing price. The data comes from the Tencent Finance market data API. The number and percentage of unlocked shares are from the aforementioned media statistics, which are from different sources and use different calibers than the daily line data.

The two narratives diverged at different points in the trading session. MiniMax-W suffered steeper declines in the morning and early afternoon before staging a recovery, while Zhipu failed to follow with a rebound in the afternoon and continued to move lower toward the close.

As of July 10	Zhipu Price / Change vs Previous Close	MiniMax-W Price / Change vs Previous Close	The Side with the Deeper Drop at the Time
9:31	1,946.0 / -4.23%	278.4 / -6.39%	MiniMax-W
11:32	1,790.0 / -11.91%	251.4 / -15.47%	MiniMax-W
13:42	Prices at the same timestamp not listed in this article	248.6 / -16.41%	MiniMax-W near intraday low
14:48	1,698.0 / -16.44%	276.2 / -7.13%	Zhipu
Close	1,640.0 / -19.29%	268.6 / -9.68%	Zhipu

This table illustrates the sequence of price movements, but does not establish causality. Unless there is verifiable evidence with timestamps, such as orders or the spread of news, a specific intraday turning point cannot be attributed to financing news, nor can actual reductions in holdings be inferred from the scale of unlocked shares.

The Price of New Money, and the Price of Old Money

MiniMax’s July 10 announcement is for a set of transactions yet to be completed: a proposed placement of 35.6 million new Class A shares, plus a HK$6.5 billion zero-coupon guaranteed convertible bond. Zhipu’s July 9 announcement, on the other hand, is a best-effort placement of 19.78 million new H shares, with the announcement explicitly noting that the transaction may not be completed.

The IPO has been completed; the Hong Kong stock follow-on financing in the table below remains a proposed transaction. The status column is intentionally shown separately to avoid treating the estimated net proceeds from the announcement as the existing cash balance.

Company	Item	Status	Price and Amount	Use of Proceeds
MiniMax-W	IPO	Completed	Offer price HK$165; gross/net proceeds from the base offering approximately HK$4.8176 billion / HK$4.5961 billion; net proceeds approximately HK$5.29339 billion assuming full exercise of the over-allotment option	R&D 90%, working capital 10%
Zhipu	IPO	Completed	Offer price HK$116.20; gross/net proceeds from the base offering approximately HK$4.3481 billion / HK$4.1734 billion; net proceeds approximately HK$4.8962 billion assuming full exercise of the over-allotment option	General-purpose large model R&D 70%, MaaS 10%, partner network and strategic investments 10%, working capital 10%
MiniMax-W	New Class A Share Placement	Proposed, subject to conditions	35.6 million shares at HK$268 per share; gross/net proceeds approximately HK$9.5408 billion / HK$9.49141 billion	Of the net proceeds combined with the convertible bonds, 80% is intended for AI infrastructure and model R&D, 10% for global expansion, 10% for working capital and general purposes
MiniMax-W	Zero-Coupon Guaranteed Convertible Bonds	Proposed, subject to conditions	Principal HK$6.5 billion, maturing in 2027; initial conversion price HK$335; net proceeds approximately HK$6.4658 billion; approximately 19.403 million new shares on a full-conversion scenario	Same as above
Zhipu (the HKEX announcement entity is Knowledge Atlas)	New H Share Best-Efforts Placement	Proposed, may not be completed	19.78 million shares at HK$1,588 per share; gross/net proceeds approximately HK$31.41064 billion / HK$31.37495 billion	Model R&D talent, computing power and related technical services; business expansion and strategic investments; capital structure and general working capital. Planned to be used by the end of 2027

If both MiniMax transactions are completed, the combined gross amount will be approximately HK$16.0408 billion, and the net amount will be approximately HK$15.9572 billion. This figure is roughly three times its IPO net proceeds after the full exercise of the over-allotment option. Zhipu’s proposed placement net amount is approximately 6.4 times its IPO net proceeds on the same basis. The numbers are large, but they do not automatically carry positive implications: money can extend the investment cycle, but it also makes the future cash return per share a more stringent question.

The placement price reflects the conditions under which the financing was negotiated, rather than the company’s overall valuation or target price. MiniMax-W’s placement price of HK$268 represents a 9.89% discount to the closing price of HK$297.4 on July 9; Zhipu’s placement price of HK$1,588 represents a 12.99% discount to the closing price of HK$1,825 on July 8. Differences in announcement disclosures, comparison benchmarks, and transaction methods mean this cannot be used to compare “which company received a steeper discount from the market.”

Convertible bonds also carry an additional layer of debt attributes, maturity, and conversion options. The initial conversion price of HKD 335 cannot be placed alongside the ordinary share placement price and treated directly as the market’s valuation conclusion for MiniMax. The secondary market closing price is merely the marginal transaction price under that day’s circulating supply, risk appetite, and information set — it does not represent the overall pricing the company obtained at the time of financing.

Implied equity value with similar IPO timing, later placed into different deal structures

Based on the IPO offering price multiplied by the number of shares issued after the IPO, the implied equity value of MiniMax is approximately HK$51.75 billion, while that of Zhipu is approximately HK$51.80 billion. This calculation merely reflects the equity prices derived from the two offerings at a unified IPO time point, and cannot serve as a substitute for subsequent market capitalization, nor should it be regarded as an official post-money valuation.

These two closely matched starting points help frame this week’s refinancing. The package offered by MiniMax is a combination of common stock and convertible bonds: one portion forms new shares after closing and issuance, while the other portion starts as debt, with future conversion depending on the terms and the holder’s choice. Zhipu is a best-efforts placement of a large amount of common stock, where the uncertainty around the new shares being issued and the completion of the financing is more directly placed on the trading table.

The market’s response on such announcement days isn’t simply answering the question of “whether anyone is willing to put up money.” It also weighs simultaneously the newly tradable or potential shares, the discount on the financing, whether the financing completes smoothly, the speed of capital deployment, and the demand for liquidity from existing shareholders after the lock-up expires. MiniMax faces a significantly larger potential supply of floating shares; Zhipu, meanwhile, confronts a best-efforts placement whose size far exceeds its IPO. These are testable backgrounds that explain the divergence—not a single-factor attribution for the moves on July 10.

There is still one more round for A-shares, but the amount cannot be factored in in advance

Both companies have disclosed the proposed RMB share matters, with varying progress and levels of information.

Company	Disclosed Matters	What Can Now Be Confirmed
MiniMax-W	Announcement on May 31 of proposed issuance of RMB shares	Scale, amount, use, or completion timeline have not been disclosed, and it is explicitly stated that implementation is not guaranteed
Zhipu	Announcement on June 1 of proposed issuance of no more than 38,768,964 new A-shares on the STAR Market, excluding over-allotment	Pricing is undetermined and no formal agreement has been reached; the RMB 15 billion represents the total planned project investment, not funds already raised. Of this, RMB 12 billion is for the general-purpose foundation large model, RMB 2 billion for MaaS, and RMB 1 billion for supplementing working capital

Adding the 15 billion yuan from this table to the cash Zhipu already holds is a common misreading. It is a project investment plan; whether the offering can actually proceed, at what price, and how much is ultimately raised still depends on subsequent procedures and disclosures. MiniMax’s announcement information is at an even earlier stage, and has not even provided the planned amount or use of proceeds.

LLM Funding Burns On — What Is the Market Watching Now?

The use of funds by large model companies is highly similar: models, compute, infrastructure, products, and globalization. The same destinations do not mean the same output. At this stage, the market’s judgment on financing has at least four sequential layers:

Whether the transaction can be completed and what the actual net amount is;
How the cash balance and future commitments on compute, training, and inference will change;
How newly issued shares and potential convertible shares alter per-share equity;
Whether revenue, gross margin, cash burn, and unit compute efficiency improve.

Refinancing allows the company to continue betting on next-generation models, and it also pushes the validation cycle to the next financial report and the next capital structure disclosure. The mid-session seesaw followed by a same-direction decline at the close on July 10 is only one slice. The true repricing will not get an answer until cash turns into sustainable revenue.

This article does not constitute investment advice.

References

写作附记

This article is written based on Hong Kong Stock Exchange announcements, market data interfaces, and clearly labeled media statistics. Refinancing and A-share matters are handled according to the proposed status at the time of announcement; lock-up release statistics are not equivalent to actual selling, and intraday price paths do not constitute news-causality proof.

Original Prompt

Zhipu and Minimax faced their first lock-up expiry wave last week. Previous articles have covered related topics, but when the moment actually arrived, the two stocks performed quite differently — Zhipu rose while Minimax fell. We will analyze why. Subsequently, financing news came out; intraday, Minimax still declined more, but by the close, Zhipu suffered a sharp drop. A table summarizes the relevant market data.

Reviewing the IPO financing scale, with the newly announced financing amounts this week, we examine how the market is pricing these two companies. There is still a subsequent round of A-share financing, along with the intended use of funds raised by both companies. With large model financing continuing to burn cash, we look at how the market views this.

The AI spending bull arrives at the lockup expiration

Sun, 21 Jun 2026 21:10:00 +0800

The awkward part of this round of AI stocks is that, while some people are shouting “capex bull,” many A-shares and Hong Kong stocks have not risen at all, and some have even dropped significantly. It looks like a contradiction, but it’s actually two sides of the same coin: money is still flooding into AI infrastructure and a handful of application gateways, but the secondary market is starting to ask who will ultimately be able to turn this spending into cash flow.

Let me directly answer your questions first.

This article can be read along three threads: North American tech giants aren’t without growth drivers—they’re betting on the next curve with heavy capital expenditure; coding is currently the easiest AI application to monetize; the key question for Zhipu and MiniMax isn’t whether they have a story, but whether their valuations and the supply unlocked by lock-up expirations can be absorbed by the market.

After the marginal efficiency of the old growth drivers at North American tech giants declines, AI has become one of the few directions that can still sustain a narrative of hundreds of billions of dollars in capital expenditure. Cloud, advertising, search, social media, and e-commerce are all still profitable, but if market caps are to continue being priced at high multiples, there must be a next curve large enough. What makes AI special is that it can be pitched both as a new product and as a reinvestment cycle for cloud computing, chips, data centers, electricity, and enterprise software.

“Capex Bull” makes sense. What is currently priced most richly is not necessarily the companies that have already earned AI net profits, but the companies standing at the front end of hyperscaler capital expenditure, collecting the money: GPUs, ASICs, HBM, optical modules, servers, switches, data centers, power equipment, cloud resources. The problem also lies here: if capex growth outpaces the realization of final revenue, stocks will first rise into a very narrow bull market, and later use pullbacks to filter out who has real cash flow.

Is the big tech industry running out of money? My take is: it’s not that they’ve become poor, but that AI investment is so massive that even companies with strong cash flow need to optimize their balance sheets. Goldman Sachs, JPMorgan, and other institutions, in their recent reports and media coverage, have estimated hyperscaler capex for 2026 at over $700 billion; Axios reported in mid-June that Nvidia has also entered a window of large-scale bond financing, driven by the growth of debt financing for AI data centers and supply chains. Having plenty of cash and continuing to issue debt are not mutually exclusive. It’s more of a signal: this cycle is not an asset-light internet bull market, but a heavy-asset balance sheet expansion.

Are investors anxious to cash out? In part, yes. IPOs, A+H listings, lock-up expirations on the secondary market, and M&A are all paths that turn paper valuations in the primary market into liquidity. This is especially true for listings like Zhipu and MiniMax, where the post-IPO float is extremely thin and stock prices have been pushed up by index inclusions and scarcity premiums. Before and after the lock-up expiration, the core issue is not simply “is the company good or not,” but rather that an originally very small tradable supply suddenly becomes much larger, and the market must reprice the liquidity premium.

Why has the coding track become white-hot? The reason is straightforward: it offers a faster path to paid willingness than “general AI changing the world.” Developers are willing to pay monthly fees for efficiency gains, while enterprises are willing to purchase seats for code generation, testing, migration, and security review. Anthropic’s earlier official fundraising materials revealed that Claude Code’s run-rate revenue exceeds $2.5 billion, and OpenAI has also turned Codex into an enterprise-grade coding agent, emphasizing in the 2026 Gartner Magic Quadrant for Enterprise AI Coding Agents that more than 4 million people use Codex weekly. Going forward, competition will not only be about model benchmark scores, but also about repository context, permission control, CI/CD integration, enterprise compliance, inference cost, and stability.

Is Zhipu’s Hong Kong stock market capitalization reasonable? Based on the closing price of 2513.HK at HK$2,094 returned by Yahoo Finance on June 18, 2026 and a total share capital of 440,230,190 shares after listing, Zhipu’s total market capitalization is approximately HK$922 billion.¹ The company’s revenue in 2025 was RMB 724.3 million. Even without precise currency conversion, this represents a price-to-sales ratio on the order of one thousand times. It has gone beyond the realm of “somewhat expensive” and can only be explained by very distant future revenue and profits. Unless you believe Zhipu will transform from a model company into China’s AI infrastructure platform, with future revenue increasing by two orders of magnitude, the current market capitalization is difficult to support based on present financial statements.

First, compress the financing data into a narrow table:

Entity	Capital Action	Amount / Valuation
OpenAI	Official new funding round	$122B committed capital; $852B post-money
Anthropic	Series H	$65B; $965B post-money
SpaceX / Cursor	Reported stock acquisition	~$60B transaction scale
Nvidia / AI Bond Financing	Reported large bond issuance	~$20B–$25B
Zhipu (智谱)	Hong Kong IPO, advancing STAR Market proposal	IPO raised ~HK$4.3B; market cap ~HK$922B
MiniMax	Hong Kong IPO, initiated A-share tutoring	IPO net proceeds ~HK$4.596B; market cap ~HK$154B

My take is set apart from the table. OpenAI and Anthropic are no longer ordinary SaaS financings, but rather infrastructure and gateway financings; Cursor has been pushed to a $60 billion deal benchmark, indicating that programming agents have shifted from a tool competition to a strategic gateway competition; core supply-chain companies like Nvidia are also starting to extend their capital duration, suggesting that AI infrastructure financing is increasingly resembling a capital markets engineering project.

Zhipu and MiniMax represent a different problem: the secondary market has assigned them a high forward platform premium, but they still have to contend with losses, secondary financing, expansion of the float, and absorbing the overhang from lock-up expirations. The meaning of this table is not “all AI companies are safe,” but rather: financing has become part of the business model. The larger the capital, the more the market will ask two questions: whether more money can still come in, and when revenue will prove that these expenditures are not sunk costs.

Programming First Becomes a Cash Flow Entry Point

Before, everyone wanted to build general-purpose AI, because the general-purpose AI story was the biggest. Then they realized: the biggest story does not equal the fastest return. C-end chat products have traffic, but paid conversion rates, retention, computing costs, and competitive subsidies are all hard to look at; enterprise general assistants can talk about efficiency, but the delivery cycles and internal compliance are heavy; programming is different — it stands directly on the software production chain.

A developer paying an extra $20, $100, or $200 per month is easy to convince, as long as it saves them a few hours. The same applies to enterprises. What truly delivers real-world value is not the slogan of “replacing programmers,” but rather fixing bugs, migrating frameworks, writing tests, reviewing code, doing refactoring, integrating internal knowledge bases, and opening PRs. These involve clear objects of work, clear before-and-after comparisons, and a clear paying audience.

So the competition between OpenAI and Anthropic will increasingly resemble a three-layered battlefield.

The first layer is model capability. Whoever can more stably understand large repositories, long contexts, and multi-file changes will have the advantage.

The second layer is the engineering shell. Tools like Codex, Claude Code, and Cursor are essentially not just model invocations—they also have to handle permissions, terminals, sandboxes, Git, testing, logs, rollbacks, and human review. If the model is strong but the shell is poor, developers will quickly get frustrated.

The third layer is enterprise distribution. The real money lies in team seats, private code security, audit logs, IDE and CI/CD integration, and admin consoles. In the future, many companies won’t just buy a single chatbot; they’ll procure AI coding agents as R&D infrastructure.

Things are very likely heading toward consolidation along this line. Standalone tools that lack their own models, compute, or deep distribution channels will easily get squeezed from both sides: upstream model providers raising prices or throttling access, and downstream big tech players bundling features directly into IDEs, cloud platforms, and office suites. The rumor of a SpaceX deal pushing Cursor’s valuation to around $60 billion is a clear sign that coding agents are now viewed as a strategic entry point, not just small utilities.

Where Zhipu’s Market Value Comes From

Zhipu’s problem is not “whether it has an AI story.” Of course it does. Its prospectus and annual results have shown that it has models, enterprise customers, R&D investment, and a capital market identity. The issue is that the secondary market has priced it too far ahead.

By the broadest measure:

\[ \text{Crude P/S Ratio} \approx \frac{\text{Total Market Cap}}{\text{Annual Revenue}} \]

On June 18, the closing price of HK$2,094 multiplied by 440,230,190 shares gives a total market capitalization of approximately HK$922 billion. Revenue in 2025 was RMB 724.3 million. Even without getting bogged down in the HK dollar and RMB conversion, this ratio has already exceeded the explanatory range that ordinary software stocks, cloud companies, and most growth stocks can bear.

To argue that it is reasonable, one must simultaneously believe several things: Zhipu’s future revenue will not be in the billions, but in the tens or hundreds of billions; the gross margin can be squeezed out from model inference and delivery costs; the ratio of R&D investment to revenue will drop significantly; and domestic government and enterprise clients, internet companies, end users, and the developer ecosystem will be willing to entrust their budgets to it over the long term.

These things are not necessarily impossible, but this is no longer “looking at the 2025 annual report to buy the company”—it is: pricing today based on a platform position that might emerge after 2030. When the float is very small, stocks like these can rise dramatically, but once lock-up restrictions are lifted, the market will ask again: are those willing to take the shares betting on future cash flow, or are they simply buying scarce liquidity?

MiniMax’s valuation is similarly not cheap. Based on a rough calculation using the closing price of 0100.HK at HK$497.6 on June 18 and total shares outstanding of 309,255,668, the total market capitalization is approximately HK$154 billion. The company’s 2025 revenue was US$79.038 million, which converts to roughly HK$600 million. The price-to-sales ratio is likewise in the hundreds range. The only difference is that Zhipu’s latest market cap is even more exaggerated, while MiniMax’s supply shock in July was greater.

What Really Changed After the July Lift

The unlock table needs to be viewed from two perspectives: the total share capital basis and the tradable share basis. The former shows potential dilution and the supply from existing shareholders, while the latter shows how many shares the secondary market can absorb on that day. To avoid crowding a wide table together, only the changes in the circulating shares are listed here:

Company	Float Before Unlock	Float After July
Zhipu `2513.HK`	~11.74 million shares	~37.42 million shares
MiniMax `0100.HK`	~12.69 million shares	Up to ~166 million shares²

Zhipu’s total share capital after listing is 440,230,190 shares. On July 7, the lock-up period on 25,681,600 shares held by cornerstone investors will expire, accounting for approximately 5.83% of the total share capital. This will increase the theoretical tradable supply from approximately 11.74 million shares to approximately 37.42 million shares, about 3.2 times the original amount.

After its IPO, MiniMax has a total share capital of 309,255,668 shares. On July 8, 16,504,040 shares held by cornerstone investors become unlocked; the upper limit of shares released from lock-up by existing shareholders is approximately 136,997,480 shares. Together, they account for about 49.63% of the total share capital, and under the upper-limit calculation, the tradable supply jumps from approximately 12.69 million shares to around 166 million shares.

The most easily misread entry in this table is MiniMax. In its allotment results announcement, the lock-up commitments shown for multiple tranches of Class A ordinary shares held by pre-IPO shareholders are set to expire on July 8, 2026; summing them item by item from the announcement gives roughly 137 million shares, consistent with the market-reported figure of “approximately 44.29% of total shares.” Combined with the 16,504,000 shares held by cornerstone investors, this yields a theoretical supply ceiling of nearly half of the total share capital.

But the key point here is: the lifting of commitments does not mean everything can be sold that same day. A footnote in the MiniMax announcement also mentions that some Pathfinder SII shares are subject to longer lock-up periods under Rule 18C; another batch of existing shareholder shares is locked up until the earlier of “the 20th trading day after inclusion in Stock Connect” and “nine months after listing.” In other words, the actual selling pressure depends on whether multiple lock-ups overlap, the nature of the shareholders, the progress of Stock Connect inclusion, and the secondary market price. The article uses the upper-bound figure to give readers a sense of the magnitude of the supply shock — it does not mean all of these shares will be sold on the same day.

Zhipu’s July unlock is much smaller. The cornerstone investors’ 25.6816 million shares will be unlocked, expanding the already thin float, but the truly large-scale lock-up period for existing shareholders falls on January 7, 2027. Zhipu’s allotment results announcement states this clearly: under Rule 18C, there is a six-month requirement, but all existing shareholders are also subject to the applicable Chinese laws prohibiting the disposal of their holdings within 12 months after listing. In other words, July is the first float expansion for Zhipu, while next January is the bigger supply test.

List the large-scale time points separately:

Zhipu, July 7, 2026: Cornerstone investors’ lock-up expires on 25,681,600 shares, approximately 5.83% of total share capital, expanding the free float from extremely low levels.
Zhipu, January 7, 2027: Existing shareholders’ 12-month lock-up expires; this includes approximately 178 million previously listed H-shares, with additional unlisted domestic shares subject to conversion and full circulation arrangements. This represents a larger potential supply window.
MiniMax, July 8, 2026: Cornerstone investors hold 16,504,040 shares, plus existing shareholders’ release commitment cap of approximately 136,997,480 shares. July is the main impact window, but overlapping lock-ups must be deducted.
MiniMax, the earlier of the 20th trading day after inclusion in Stock Connect or October 8, 2026: Another batch of existing shareholders’ shares of approximately 56.95 million shares. Whether this occurs earlier than October depends on Stock Connect trading availability.
MiniMax, January 8, 2028: Controlling shareholders’ 24-month lock-up expires, approximately 79 million shares. This corresponds to founder and control-related holdings, with a more distant timeline.

So, if the question is who is under more pressure in July, the answer is straightforward: MiniMax. Zhipu will also expand its circulation in July, but its real test comes later. MiniMax’s problem is that the theoretical supply in July is too large. If the stock price is still in a high valuation range, early shareholders don’t need to sell all of their holdings — the market will first price in the risk that they “might sell.”

Why Have Many A-Shares and Hong Kong Stocks Not Risen

This stock market rally is not a broad-based bull market. The situation is particularly evident in A-shares and Hong Kong stocks, because many companies have not directly benefited from AI capex, and may even be weighed down by high interest rates, weak consumption, the real estate chain, local government finances, export expectations, and domestic demand expectations.

The funding path for the AI capex bull market is narrow: first buy the American large caps and the chip chain, then expand into electricity, data centers, optical communications, storage, and a few application entry points. Among A-shares and Hong Kong stocks, there are many companies with AI in their names, but there aren’t that many that can truly turn orders into revenue, revenue into gross profit, and gross profit into cash flow. Once the market shifts from “having an AI concept” to “whether it can deliver,” most stocks will be left behind.

Hong Kong stocks have an additional issue: when many new-economy companies go public, the tradable float offered is quite small, and short-term price gains can be pushed significantly higher by scarcity and index-related buying. Once lockups expire, the market finally confronts the real supply. Zhipu and MiniMax are a microcosm of this logic: inclusion in the Hang Seng Tech Index, expectations for Stock Connect eligibility, and their status as scarce AI targets all attract buying; but lockup expirations, losses, secondary financings, and elevated price-to-sales ratios cause that buying to start hesitating.

My overall assessment is: AI is not without growth drivers, nor is AI purely a bubble; however, if this bull run is primarily driven by spending, it will naturally lead to divergence. Companies that sell the picks and shovels benefit first, products that can turn AI into a high-frequency paid workflow benefit next, and companies that only talk about long-term general-purpose AI without visible cash flow will find it increasingly harder to justify their valuations.

The programming track will continue to see fierce competition going forward, because it was the earliest direction that proved “AI can make money by seat, by usage, and by team budget.” OpenAI, Anthropic, Cursor, as well as the major players’ IDE/cloud platforms will all be crowding into this space. It won’t stop at “writing a few lines of code,” but will dig into the software delivery process: requirements, code, testing, review, deployment, security, and operations can all become billable points for agents.

But the stock market won’t wait for all the answers to become clear. It will first give high valuations to the companies that most resemble infrastructure, and then gradually test them through lock-up expirations, financing, debt, quarterly reports, and pullbacks. The question now is not “does AI still have a future,” but rather: how much future has already been priced into these valuations.

This article only provides a compilation of publicly available materials and personal analysis, and does not constitute any investment advice. Market conditions, market capitalization, and lock-up release figures may change with announcements and market developments. Specific trading decisions should still be based on officially disclosed documents and your own risk constraints.

References

OpenAI: OpenAI raises $122 billion to accelerate the next phase of AI
Anthropic: Anthropic raises $65B in Series H funding at $965B post-money valuation
Anthropic: confidentially submits draft S-1 to the SEC
OpenAI: named a Leader in enterprise coding agents by Gartner
OpenAI: Codex for every role, tool, and workflow
Axios: SpaceX will buy Cursor for $60 billion
Axios: AI debt boom ramps up with Nvidia bond sale
Business Insider: Goldman Sachs on AI capex supercycle
MiniMax Group Inc. Prospectus, HKEX filing
MiniMax Group Inc. Allotment Results, HKEX filing
MiniMax Group Inc. 2025 Annual Results Announcement, HKEX filing
Beijing Zhipu Huazhang Technology Co., Ltd. Prospectus, HKEX filing
Beijing Zhipu Huazhang Technology Co., Ltd. Allotment Results, HKEX filing
Beijing Zhipu Huazhang Technology Co., Ltd. 2025 Annual Results Announcement, HKEX filing
Yahoo Finance: Historical quotes for 0100.HK, 2513.HK, ^HSI, retrieved on 2026-06-21; 0100.HK only returned one record on 2026-06-18 this time, so no complete historical series was expanded based on this.

写作附记

Original Prompt

$blog-writer I actually haven’t quite figured it out—have the major North American tech companies all run out of new growth drivers? They’re all rushing into AI, and some say this bull market is a “capex bull,” driven by investments from the big players. Analyze the funding data and the latest funding news, organize them into a table, and share your perspective on this funding data. Are the big companies themselves short on cash? Are investors also anxious to cash out? Earlier, everyone wanted to build general-purpose AI, but later they realized it still needs to make money. The coding track is profitable—OpenAI and Claude are competing, and it’s already turning white-hot. Where will things go from here? Is the market cap of the Hong Kong-listed Zhipu reasonable? In July, both MiniMax and Zhipu had a wave of lock-up expirations. Analyze the changes in floating share data before and after the lock-up expirations—when is the large-scale lock-up expiration, and what are the corresponding data changes. In this stock market cycle, many stocks haven’t risen, and some have even dropped significantly—this is especially evident in A-shares and Hong Kong stocks. As a professional financial analyst, please answer my questions first.

Writing Approach Summary

This piece retains the core questions from the original prompt: Kai-fu Lee’s expenses, funding news, the programming track, Zhipu valuation, MiniMax and Zhipu lock-up expirations, and the A/H stock divergence. What was suppressed are unverifiable absolute statements, such as “the major companies can’t find any growth points” and “investors must be anxious to cash out.”

The body of the text divides the materials into three layers: the company’s official financing and performance are first-hand facts; capex, bonds, and the Cursor transaction use recent media reports as the latest background; market cap, free float, and the unlocking schedule are rough estimates based on HKEX disclosure documents and the market data as of 2026-06-18. MiniMax’s July unlocking is specifically marked with “ceiling caliber” and “overlapping lock-up” to avoid directly writing theoretical supply as same-day selling pressure.

The Hong Kong stock price and rough market capitalization in this article are calculated using the closing price on 2026-06-18; 0100.HK only returned one quote for 2026-06-18 from Yahoo Finance this time, so the full historical price trend of MiniMax is not expanded accordingly. ↩︎
This refers to the theoretical upper-bound calculation. Certain MiniMax shareholders, including Pathfinder SII and others, may still be subject to Rule 18C, Stock Connect trading hours, or other lock-up arrangements. The lifting of the lock-up commitment does not mean that all shares will be sold on the same day. ↩︎

Xiaomi drops back to hardware ledger: Storage price hikes, auto subsidies, and AI spending hit simultaneously

Fri, 12 Jun 2026 08:20:23 +0800

Xiaomi’s current downturn cannot be attributed solely to an “AI spending spree” narrative. Over the past year, the stock price fell steadily from a high of around HKD 60 in July 2025 to HKD 25.84 on June 11, 2026. What is truly difficult is the convergence of three factors: smartphone profits are squeezed by rising storage costs; the auto sector has moved from rapid volume growth to a stage of subsidy tapering and model switching; and AI investments have pushed the market’s patience for free cash flow further into the future.

I want to start with a judgment: Xiaomi’s profit recovery doesn’t necessarily have to wait for the end of this crazy AI wave, but it must wait until the rising slope of storage and smartphone component costs slows down, or until Xiaomi can offset these costs using higher ASP, less low-end inventory, and automotive scale profits. AI is not the only issue; AI is more like an amplifier for this round of storage price increases and capital expenditure.

Stock Price First Fell Due to Profit Doubts

I pulled adjusted daily data for 1810.HK from Yahoo Finance covering June 12, 2025, to June 11, 2026. During this period, Xiaomi dropped from HKD 52.20 to HKD 25.84, a decline of 50.5%. The peak for the range was HKD 60.15 on July 2, 2025. The latest price is exactly the low point of the range, representing a 57.0% pullback from the high.

Time Point	Adjusted Closing Price	Interpretation
2025-06-12	52.20 HKD	Start point one year ago
2025-07-02	60.15 HKD	High point of the range
2025-09-30	54.00 HKD	Still on a high platform
2025-12-31	39.36 HKD	Valuation has clearly dropped by year-end
2026-03-31	31.76 HKD	Q1 performance pressure starts to be factored in
2026-06-11	25.84 HKD	Low point of the range

There is noise in the capital side of things. The RMB has indeed appreciated over the past year; Yahoo Finance’s USDCNY=X dropped from 7.1928 to 6.7725, representing a decline of 5.84% for USD against RMB. HKD/RMB also fell from 0.9164 to 0.8643. When viewing Hong Kong stocks with RMB, the same HKD assets will appear more expensive, and southbound funds may experience currency exchange rate fluctuations and portfolio pacing in the short term.

But framing this as “RMB appreciation leads to decreased southbound capital, which causes Xiaomi to fall” is too convenient. HKEX’s Stock Connect 2025 Review shows that the average daily turnover for southbound funds in 2025 was HKD 121.1 billion, compared to HKD 48.2 billion in 2024; at the end of Q4 2025, southbound turnover accounted for 23.0% of Hong Kong cash stock turnover, which is higher than the 20.9% recorded at the end of Q4 2024. HKEX monthly data also shows that in the first five months of 2026, Hong Kong stocks averaged a daily turnover of HKD 275.3 billion, representing a year-on-year

A more direct explanation can be found in the income statement. Xiaomi’s full-year revenue in 2025 was 457.3 billion yuan, with smart EVs, AI, and other new businesses contributing 106.1 billion yuan in revenue. Operating profit for the entire year turned positive for the first time at 0.9 billion yuan, and they delivered 411,082 vehicles. Looking only at the full year, the narrative is not bad. However, by Q1 2026, the reported figures became tighter: revenue was 99.1 billion yuan, adjusted net profit was 6.1 billion yuan, a year-on-year decrease of approximately 43

The stock price drop isn’t due to “Xiaomi not making cars” or “Xiaomi lacking AI,” but rather because the market has started asking an old question again: Can Xiaomi truly protect its profit margins amid fierce hardware competition?

Several Headwinds Are Not the Same Kind of Pressure

The downside risks listed in the user prompt can be divided into three layers and should not be mixed together.

The first layer is valuation and capital discounting. US dollar liquidity, overseas capital risk appetite, and the overall trading rhythm of Hong Kong stocks all affect high-beta HK stocks like Xiaomi. The Fed’s balance sheet will still be around $6.7 trillion until June 2026, and the market continues to discuss smaller balance sheets and reserve requirements. Naturally, foreign capital will not provide infinite duration/support for growth stocks as it did during the low-interest rate era. However, this type of pressure is more related to a “valuation multiple” issue; it typically cuts P/E or P/S first, rather than directly changing how much money Xiaomi earns from a single phone.

The second layer is automotive subsidies and EV competition. For 2024-2025, new energy passenger vehicles will be exempt from purchase tax, with a single vehicle exemption cap of ¥30,000; for 2026-2027, this will change to half the rate, with a single vehicle tax reduction cap of ¥15,000. For vehicles priced between ¥200,000 and ¥300,000, this is no small amount. It will have two consequences: it will accelerate some demand before the end of 2025, and in 2026, car manufacturers must either absorb some price pressure or accept a slower order rhythm. Xiaomi Auto proved its ability to increase volume in 2025, but if it delivers 80,856 units in Q1 2026—a drop of 44.3% compared to 145,115 units in Q4 2025—the market will no longer solely focus on the “full-year target of 550,000 vehicles,” but will also monitor model switching and per-vehicle profit.

The third factor is mobile phone costs. This pressure is the strongest because it is already reflected in the gross profit margin. In Q4 2025, Xiaomi’s mobile phone gross profit margin dropped to 8.3%, and in Q1 2026, it was only 10.1%. During this period, phone shipments were 33.8 million units, marking a year-on-year decline of about 19%. However, the ASP actually rose to around 1,310 RMB. This indicates that Xiaomi is making an uncomfortable adjustment: reducing low-end and mid-low-end models that are under significant pressure, and moving towards higher price points. But even so, the gross margin continues to be eroded by cost and competition.

The role of AI here is a bit complex. Xiaomi itself plans to invest in AI, intelligent driving, and embodied intelligence. Caixin’s summary mentions an investment budget of 16 billion RMB for AI and embodied intelligence in 2026. The bigger issue is that global cloud vendors and server demands are drawing the capacity of DRAM, HBM, and NAND towards data centers, causing smartphones, PCs, and consumer SSDs to start queueing on the same capacity chart. Xiaomi is not paying solely for “doing AI for itself,” but rather contributing to the competitive bidding for the entire industry’s AI infrastructure.

How to Buy a Phone Amid Rising Storage Costs

Over the past two years, storage prices have not risen smoothly; rather, they first recovered and then saw an uncontrolled surge.

I kept HDD separate in the table because users asked about mechanical hard drives. However, it has no direct BOM impact on Xiaomi phones. HDD affects Xiaomi notebooks, NAS/ecosystem chain, cloud AI storage, and the entire consumer electronics price environment. What actually goes into Xiaomi phones are LPDDR4X/LPDDR5X and eMMC/UFS.

Taking several model types still displayed on the global product pages of the Xiaomi official website as an example, the cost pressures are roughly as follows:

Model Sample	Official Website Storage Configuration	Rough Benchmark for RAM + Flash in 2025	Added Cost Pressure in H1 2026	Explanation
Redmi 14C	4/128 to 8/256, LPDDR4X + eMMC 5.1	Approx $8-$12 USD	Approx +$4 to +$8 USD, approx ¥30-¥60	In the Counterpoint Q1 2025 Top 10 best-selling list, Redmi 14C 4G is the only model listed besides Apple and Samsung; it is the hardest type of budget phone to raise prices for.
Redmi Note 15 Pro+ 5G	8/256, 12/256, 12/512, LPDDR4X + UFS 2.2	Approx $20-$28 USD	Approx +$10 to +$18 USD, approx ¥70-¥130	Mid-range phones can build configurations using cameras, screens, and batteries, but a cost increase of the ¥100 level might clash with promotional prices and channel margins.
Xiaomi 15 Ultra	16/512, 16/1

This table is neither a disassembled BOM nor Xiaomi’s procurement contract. It merely performs a stress test using public specifications and industry price increase ranges. Actual procurement prices are influenced by long-term agreements, inventory cycles, vendor rebates, currency, model types, and delivery windows. However, even under conservative assumptions, the conclusion is clear: entry-level phones face increases of tens of yuan, mid-range phones an increase of around a hundred yuan, and high-end phones several hundred yuan. This is no small matter for a company that relies on scale and high cost-performance to ship products.

This also explains why it is impossible for Xiaomi to simply pass on every cost increase through price hikes to consumers. For a volume model like the Redmi 14C, raising the price

Focus on the Substance, Not the Slogans

Therefore, Xiaomi’s profit recovery does not need to wait for AI to completely fade away. There are three more realistic trigger points.

The first point is the slowdown in the slope of storage prices. As long as DRAM and NAND maintain their rhythm of quarterly explosive growth (like observed in Q1 or Q2), mobile phone gross margins will struggle to be sustainable. Even if prices do not return to 2024 lows, as long as the rate of increase drops from “near quarterly doubling” to a normal cycle price hike, Xiaomi can gradually repair its position through inventory management, product mix adjustments, and pricing strategies.

The second point addresses whether raising the mobile ASP (Average Selling Price) has negatively impacted sales volume. A contradiction observed in Q1 2026 is that while the ASP reached a new high, shipments saw a year-on-year decline of approximately 19%. If the company merely increases prices to maintain gross profit margins going forward, stock prices may not react positively; only if Xiaomi can stabilize the ASP using both high-end and core mid-range models, while simultaneously narrowing the decline in overall shipments, will the market revalue the mobile business.

The third point is whether the automotive business can once again prove its scale profitability. Throughout 2025, EV, AI, and other new businesses achieved positive operating profits, but Q1 2026 incurred a loss of 3.1 billion RMB. This may not necessarily be a bad thing; model transitions, production pacing, and pricing strategies can all cause quarterly fluctuations. The key issue is that after the subsidies taper in 2026, Xiaomi must prove that the YU7/SU7 are not solely reliant on order hype, but can maintain gross margins and deliveries even in a more normal subsidy environment.

Regarding the stock price movement, the vicinity of HK$25 is already the low point from the past year, and the technical structure remains weak. My baseline judgment is that without new evidence of profit repair, the short term is more likely a weak rebound within the HK$25–32 range; if HK$25 breaks down, the market will look for the next layer of support in the HK$22–24 range; and only if subsequent financial reports confirm that phone gross margins have stabilized and auto deliveries reach a new plateau should we first watch HK$32 before having an opportunity to challenge the HK$35–40 region. This range is not a buy/sell recommendation, but merely an observational framework based on combining current profit pressure and the price distribution over the past year.

I am rather reluctant to base my judgment solely on the premise of “waiting for the AI frenzy to end.” If the AI boom completely fades, storage prices might moderate, but Xiaomi’s narratives around its cars, mobile AI, intelligent driving, and high-end devices will also lose some valuation elasticity. A better state for Xiaomi is not the bursting of the AI bubble, but rather the transition of AI infrastructure from a panic buying stage to a stable procurement phase, thereby preventing storage prices from rewriting the hardware ledger on a quarterly basis.

References

Writing Notes/Epilogue

Original Prompt

$blog-writer Xiaomi's stock price faces multiple headwinds: the appreciation of the RMB caused major Southbound funds to reduce investments in Hong Kong; tightening USD liquidity also reduced foreign investment capital. Furthermore, electric vehicle subsidies are expiring, national subsidies are tapering off, and AI continues to burn money. Storage chip prices rising, coupled with AI hype, leads to continuously increasing semiconductor prices. Will Xiaomi's profit recovery only occur after the current AI frenzy completely subsides? Compile and analyze the contents I mentioned earlier, gathering Xiaomi's stock price data from the past year to predict its subsequent trend. Regarding semiconductor price increases, compile the annual inflation rates for Solid State Drives (SSDs), Hard Disk Drives (HDDs), and memory over the past two years, and look up Xiaomi's popular models to assess how much the machine's hardware costs have increased.

Summary of Conceptual Ideas

This article retains the several themes proposed by the user: funds, subsidies, AI, storage, and stock prices, but removes the phrasing that every downside risk develops equally. RMB appreciation and Southbound funds are downgraded to funding rhythm factors because HKEX data does not support portraying the entire southbound flow as distinctly shrinking. Storage price increases and mobile phone gross profit margins are stronger main themes, so the body text allocates more space to cost breakdown analysis.

The cost estimation only provides a range, not a specific Xiaomi purchase price. The actual purchasing price requires information such as contracts, inventory levels, suppliers, and delivery windows, which cannot be confirmed using public materials.

After an AI stock pullback, first look at how the crowded trades unravel/dissipate.

Sun, 07 Jun 2026 04:24:52 +0800

Looking back at this on Beijing Time June 9, 2026—the line from June 5 is no longer sufficient. On June 8, the A-share market failed to digest last Friday’s retracement of tech stocks; instead, the ChiNext board, STAR 50, CPO, and semiconductors continued to decline sharply. Simultaneously, US stocks also saw intraday pullbacks on June 8, with QQQ, SOXX, and a group of AI chip stocks exhibiting obvious rebounds. When looking at both markets together, the conclusion is even less straightforward: A-shares are continuing to dismantle crowded trades, while US stocks are performing high-elasticity rallies intraday. Neither of these events serves as concrete evidence that “clearing has been completed.”

First, let’s clarify the timeline. A-shares fell during the day on June 5th, mainly due to issues with their own high-flying tech stocks, financing constraints, and sector overcrowding; US stocks declined on the evening of June 5th, pressured by a combination of strong nonfarm reports, yield concerns, and chip stock expectations. The continued slide in A-shares on June 8th suggests further pricing for crowded trades after Friday’s pressure. The intra-day rebound in US stocks on June 8th only indicates that the AI hardware chain—which saw the most selling on the previous trading day—still has funds replenishing; this cannot be prematurely stated as a closing conclusion.

On A-shares, on June 5th, the SSE Composite Index

Up to June 8th, the pressure didn’t stop at “a single day of falling.” According to the closing figures from Daily Economy News, the Shanghai Index fell 1.7%, the Shenzhen Component Index fell 3.22%, the ChiNext Index fell 3.69%, and the STAR 50 Index fell 4.30%, with nearly 4,600 stocks closing lower. The Securities Times also reported on the same day that the three main markets (Shanghai, Shenzhen, and Beijing) transacted a combined total of about 2.82 trillion yuan, while the BEST 50 index rose 1.33% against the trend. On the market board, CPO and semiconductor chips continued to lead declines, while sectors like banking, oil & gas, coal, and robotics were relatively stronger. This combination was more akin to a structural rebalancing of trades than on Friday: high-flying technology stocks continued shedding excess gains, while low-lying or defensive assets absorbed some risk appetite.

In the US stock market, the decline on June 5th was more pronounced. According to AP’s closing data, the S&P 500 fell by 2.6%, the Dow Jones Industrial Average dropped about 1.3%, and the Nasdaq fell 4.2%. The BLS employment report released the same day showed that nonfarm payrolls increased by 172,000, and the unemployment rate remained at 4.3%. While this data was stronger than market expectations, the result wasn’t “good economy means good stocks,” but rather rising yields and delayed interest rate cut expectations, which caused a re-discounting of future cash flows for highly valued tech stocks. Axios described this day as the decline in the Nasdaq driven by chip stocks, while reports cited by Reuters mentioned that US chip companies lost over $1 trillion in market capitalization on that day.

However, by midday on June 8, 2026, Eastern Time, the market was not continuing in a one-directional decline. According to the chart tool’s data during the trading session around UTC 16:21, QQQ rose approximately 2.36%, SPY rose approximately 0.86%, and DIA rose approximately 0.24%; SOXX rose about 7.27%. Among individual stocks, Nvidia rose by approximately 2.26%, Broadcom rose by approximately 3.14%, AMD rose by approximately 5.27%, Marvell rose by approximately 13.95%, Micron rose by approximately 11.11%, and Taiwan Semiconductor ADR rose by approximately 3.73%. I am only analyzing these figures based on the midday reading, as US stocks have not closed yet; a midday rebound is not the same thing as confirmation of funds at closing.

The most easily misinterpreted statement here is, “Once the domestic market drops, followed by a drop in US stocks, a US rebound means AI is safe.” I do not view it that way.

On Friday, both sides shared pressures related to valuation and trading structure; on Monday, A-shares confirmed that crowded trades are still dissipating; and the intraday rebound of US stocks confirmed that there is still capital willing to make up losses in high-elasticity assets.

There is a correlation among the three, but they do not follow the same single causal line.

A-Shares Continue to Decongest

I will analyze the main drivers behind the A-share market over the last two days across three levels/perspectives.

The primary market observation is this: the sectors experiencing the largest declines, and contributing most to the index drag, were not niche, low-priced assets, but rather segments like semiconductors, storage, optical modules, computing power hardware, and sci-tech weights—sectors previously heavily concentrated by funds. Furthermore, the ChiNext Board and STAR 50 performing consistently weaker than the Shanghai Composite Index themselves indicate that growth stocks and technology beta are under selling pressure.

The second layer is the trading structure. During an after-hours interview conducted by the Twenty-First Century Business Report on June 5th, fund managers cited “multiple brokerage firms lowering the financing conversion ratio for leading semiconductor and AI companies” as a major catalyst. I treat this statement only as one market interpretation, not an official conclusion. However, it explains why high-flyer stocks fall more steeply: When a trade is too crowded, any variable affecting leverage, margin, conversion ratio, and financing capacity tends to be amplified more easily than usual.

The third layer is style rotation. The fact that not only tech stocks fell on June 8th, but nearly 4600 stocks closed lower indicates that risk appetite has broadened; however, the resilience of sectors like banking, oil/gas, coal, and the BEST 50 suggests that funds have not dumped all risk assets indiscriminately. Instead, it appears to be a withdrawal from the high-flyer tech grouping—with some capital shifting into defense plays, while another portion seeks relative strength (or momentum) in small caps and other themes.

Therefore, I prefer to call this A-share decline a concentrated unwinding of high-flyer tech stocks, rather than a comprehensive macro-level risk purge. This difference is crucial. The former implies divergence within sectors, while the latter suggests all risk assets must be treated as if they were undergoing a systemic crisis. The additional information as of June 8th is that the unwinding has not finished and has transitioned from being a “Friday single-day shock” to one that requires confirmation on subsequent trading days.

Mid-day Rebound in US Stocks Does Not Signal Cleansing Completion

The pressure on U.S. stocks this Friday feels like two buttons being pushed at the same time.

One issue is interest rates. The BLS’s non-farm data itself isn’t bad, but for high-valuation tech stocks, overly strong employment data makes the market worry that the Federal Reserve will find it harder to pivot toward easing. AI stocks are especially susceptible to this because much of their valuation comes from revenues and profits many years into the future. Even a slight increase in the discount rate will cut a significant chunk out of long-term stories.

Another factor is expectation. Broadcom’s official financial report is not a sign of failure for an underperforming company. It reported that its Q2 AI semiconductor revenue for fiscal year 2026 reached $10.8 billion, representing a 143% year-over-year increase, and guided total revenue for Q3 to approximately $29.4 billion. The issue is that AI chip stocks have already risen too much; the market doesn’t want “good,” it wants something that exceeds the already highly elevated expectations. When the earnings report cannot continue to raise that imagined curve, capital will start selling immediately.

The mid-day rebound on June 8th cannot be dismissed. The faster assets like SOXX, Marvell, Micron, and AMD rebound, the more it demonstrates that they remain cores of high volatility trading. Short-term capital tends to chase rebounds after a sharp decline, but what truly determines whether this withdrawal cycle has ended is not the mid-day gains, but rather whether the market can maintain stability at close; whether volume shifts from panic profit-taking to steady buying interest; and whether interest rate expectations will continue to suppress valuations.

This is the strangest thing about AI stocks: while industry trends remain strong, the stock can plummet severely; yet, after a sharp fall, it can rebound quickly. This is because stock trading isn’t about “whether AI has a future,” but rather how much future value has already been priced into this current price, and how many other people in the market are buying using the same rationale, leverage, and timeframe.

By consolidating the A-share data from June 5th and June 8th with the US stock intraday data from June 8th, what truly needs to be examined are not directional slogans or narratives, but these variables:

See three lines below

Moving forward, I will avoid an “either/or” judgment such as “buying the AI dip” or “avoiding AI.” AI remains one of the strongest core themes in capital expenditure and industrial upgrading over the next few years, but following drawdowns like those on June 5th and June 8th, the market will become more discerning.

The first point is whether the crowdedness has truly dispersed. For A-shares, we need to see if high-turnover individual stocks in semiconductors, optical modules, and computing power hardware continue falling on increased volume, or if they start stabilizing with decreased volume; we must also determine whether the strength seen in banking, oil/gas, coal, and small-cap themes is merely temporary safe-haven buying, or if funds are continuously reducing their tech allocation. For US stocks, we need to see if the Nasdaq, SOXX, and AI chip stocks can maintain their intraday rebound after market close, rather than basing our judgment solely on midday quotes.

The second focus area is monitoring how cash flow from AI investments is realized. In the hardware chain, the most valuable asset is not simply having “AI” in your name, but rather whether orders can translate into revenue, revenue into gross profit, and gross profit into free cash flow. For many A-share companies mapping this sector, it is more crucial to examine customer concentration, delivery pace, expansion capital expenditure (CapEx), and accounts receivable; while for US stocks, the focus should be on whether cloud vendors’ CapEx has slowed down, and if guidance from Nvidia, Broadcom, AMD, storage, and optical communication chains continues to be revised upwards.

The third factor is whether valuation remains sensitive to interest rates. If US employment and inflation data continue to push back rate cut expectations, highly valued AI stocks will repeatedly face pressure from the discount rate. This pressure might not change the industry trend, but it will change the stock rhythm: previously the market was willing to pay for 2028 profits; now, it may only be willing to pay for visible orders in 2026 and 2027.

This consequently leads me to categorize the AI stocks into three types when analyzing them, rather than grouping them all together (or: putting them all in one basket).

One type consists of infrastructure companies that truly possess pricing power, strong cash flow, and a locked-in customer base. They will also decline, but once they bottom out, it will be easier for the market to re-evaluate them.

One type consists of companies with high elasticity, supported by industry chain mapping and order expectations. While their gains are rapid, they also experience larger drawdowns, and they are particularly susceptible to funding constraints, disappointing performance results, and rumors of customers cutting orders.

There is another category: companies that rely purely on concept diffusion. In a bull market, they appear the most speculative, and during drawdowns, it is also hardest to find fundamental support points. Following June 5th and June 8th, this type requires a higher risk discount because the market has begun shifting from “narrative tolerance” to “realization scrutiny.”

My conclusion is limited: this pullback does not represent the end of the AI industry trend, but it is highly likely a reminder regarding how AI stocks are valued. Going forward, when looking at AI, we cannot simply ask, “Is it AI?” We must instead ask: what part of the money spent on AI capital expenditures (CapEx) will this company actually capture? Can they maintain their gross margins? When will cash flow return? And has the current valuation already priced in the good news from the next two or three years?

If these questions cannot be answered, a dip might not necessarily be cheap; but if they can be answered, a large drop will merely wash out trend-following funds. What is more difficult now is that the mid-day rebound in US stocks gives people a false sense of “everything is fine,” while the A-share close on June 8th reminds us that when crowded trades truly dissipate, it is usually not so graceful.

This article only presents a market review and personal observations, and does not constitute any investment advice. The data parameters for single-day and intra-day market movements may vary slightly depending on the exchange, charting software, or media statistical methods. Specifically, the US stock data for June 8th was not at the closing price/official close when this text was written. Specific trading decisions should always be based on one’s own account constraints, risk tolerance, and officially disclosed information.

References

Artificial Intelligence Machine Learning

Data Mining
Deep Learning
Neural Network

Original Prompt

$blog-writer Last Friday, China's A-share market experienced a sharp drop. Analyze the main reasons for the decline. On Friday night, US stocks also plummeted. What is the outlook for AI-related stocks afterward?

This Update’s Prompts

Today is June 9th. Based on the A-share market data from June 8th, and the current gain/loss percentage of US stocks (not yet closed), I'm updating the article "07-AI-After Stock Pullback - First Look at How Overcrowded Trades Dissipate".

Summary of Writing Ideas

This update maintains the original core judgment: A-shares and US stocks cannot be forced into a single causal narrative, but both exposed issues regarding AI trade crowding, interest rate sensitivity, and earnings realization pressure. The newly added A-share data for June 8 shifts the focus from “single-day drawdown on Friday” to “potential continued overtrading/crowding on the following trading day”; meanwhile, the newly added intraday US market data reminds readers not to treat the midday rebound in semiconductor stocks as if the cleansing process is complete.

I have omitted individual A-share stock gains/losses and intraday trading advice because such content tends to turn a retrospective analysis into short-term instructions. The main body only retains indices, sectors, trading volume, and intraday US market fluctuations that can support the judgment, while repeatedly marking the boundary of “US stocks not yet closed” in relevant paragraphs.

The renewed surge of A-share semiconductors should not be bought based on industrial logic alone; caution is needed as this may constitute a crowded trade.

Mon, 25 May 2026 22:00:21 +0800

When the A-share semiconductor and AI hardware chains surge rapidly, two extreme reactions tend to emerge: one type of person feels that missing out on gains (FOMO) is more painful than actually incurring losses, while another believes that excessive rises necessarily signal a bubble.

Both of these are too fast. In areas like semiconductors, computing power, optical modules, and storage, the industrial logic might be true. AI training and inference will indeed boost hardware demand, and domestic substitution has certainly provided narrative space and order opportunities for local companies. The problem is that just because the industrial logic holds true does not mean that the probability of investing in it now is good.

Similar market trends have occurred repeatedly in history: the liquor sector (Baijiu), new energy, pharmaceuticals, core asset grouping, and TMT. Each time, there was real logic behind it. When these sectors decline, it doesn’t necessarily mean the logic has disappeared; rather, the timing/rhythm between valuation, positioning, earnings realization, and liquidity was off.

Let’s first see how much it has risen in this round

Let’s first clarify the focus. The strongest “AI-related companies” on A-shares right now are not the general media concepts from 2023, but rather three more solid value chains:

Semiconductor manufacturing and equipment agency.
Independent CPU / GPU / AI chips.
Optical modules, optical components, and compute infrastructure.

The gains in the table below are uniformly calculated using Yahoo Finance’s fully adjusted daily data, as of the close on 2026-05-25.

Proxy or Company	Code	Period	Period Gain (%)
半导体 ETF 代理	`159995.SZ`	`2026-04-07` 到 `2026-05-25`	`66.1%`
AI ETF 代理	`515070.SS`	`2026-04-07` 到 `2026-05-25`	`40.5%`
中芯国际	`688981.SS`	`2026-04-07` 到 `2026-05-25`	`63.1%`
海光信息	`688041.SS`	`2026-04-07` 到 `2026-05-25`	`51.5%`
寒武纪	`688256.SS`	`2026-04-07` 到 `2026-05-25`	`87.5%`
中际旭创	`300308.SZ`	`2026-04-07` 到 `2026-05-25`	`76.6%`
新易盛	`300502.SZ`	`2026-04-07` 到 `2026-05-25`	`43.7%`

If we only look at the slope over a little more than a month, this is no longer a “slow bull recovery,” but rather a highly typical primary rally driven by the resonance between sentiment and fundamentals. The question isn’t whether its gains are justified, but how much of that narrative has already been priced into the stock.

It’s Not Just AI: Many Concepts of “True Logic” Have Gone Through Their Failures

To summarize: Market cycles like this have happened before, and they are not limited only to tech sectors such as AI, semiconductors, or TMT. The more common scenario is that the underlying logic remains, the cyclical peak hasn’t immediately ended, but the steepest part of the price action has already priced in (or absorbed) the optimistic expectations for the next few quarters, or even years.

The logic behind Baijiu was not fake; the leading brands’ cash flows and brand moats were genuine. However, buying consumption certainty at valuations of 50x or 60x meant that repayment was still required later on.
New energy is no exception to this rule either. Although the penetration rate continues to rise, when supply expands, price wars erupt, and profits decline all at once, stock prices will not wait for the long-term narrative to slowly materialize.
Pharmaceuticals and medical services are even more typical. Many companies’ underlying businesses remain strong

So, today’s semiconductor and AI hardware chains certainly have genuine substance. However, the cluster of baijiu, new energy, medicine, and blue chips also once had real potential. The subsequent chaos wasn’t because all the initial narratives were fraudulent schemes, but rather because prices turned “the future will be good” into “the future must be perfect.”

With this batch of companies now, are their earnings supporting the stock price?

What truly needs to be looked at is not “whether semiconductors have a future,” but whether “the growth rate of profits can keep up with the slope of the stock price from the last six or seven weeks.” I will break down several representative companies:

Company	Industry Chain	2025 Revenue / YoY	2025 Net Profit Attributable to Parent / YoY	2026 Q1 Revenue / YoY	2026 Q1 Net Profit Attributable to Parent / YoY	Stock Price Increase from `2026-04-07` to `2026-05-25`	My Key Focus Area
SMIC (Semiconductor Manufacturing International Corp)	Wafer Fabrication	`67.323` billion yuan, `+16.49%`	`5.041` billion yuan, `+36.29%`	`17.617` billion yuan, `+8.1%`	`1.361` billion yuan, `+0.4%`	`63.1%`	The stock price slope is significantly faster than the quarterly profit growth rate,

This table contains two important signals.

First, not all major rallies are driven by short covering plays. This is especially true for companies like Zhongji Xuchuang, Xin Yisheng, and Cambricon; this current cycle is not purely thematic momentum—profits and orders are genuinely being realized. Because of this, many people naturally become complacent about potential pullbacks, thinking that “if there are solid earnings, a deep correction won’t happen.”

Second, earnings realization does not guarantee a safe buy-in point. SMIC and Seagull Information have clearly demonstrated this point: although their industry logic is completely robust, their short-term stock price momentum has significantly outpaced the latest quarterly profit growth rate. The market doesn’t just look at this current report; it looks at whether you can maximize (or deliver) the market share, ASP, capital expenditure, and domestic substitution potential for the coming years.

This is also why, even when experiencing a “true boom” (真景气), some people can profit from the trend, while others get washed out during pullbacks/drawdowns. What was wrong was not the direction, but the timing.

Increasing Semiconductor Positions Now: The issue isn’t whether it is right or wrong, but the timing (position).

If I had to summarize it in a single sentence, my judgment is:

Adding to semiconductor positions now isn’t necessarily that the direction is wrong; rather, the risk-reward ratio is no longer as favorable as it was in early April. This is especially true for new capital—it looks more like chasing momentum from high levels rather than low-risk positioning.

I will divide the current semiconductor and AI hardware chain into three categories:

Group	Representative Company	Current Strongest Support	Current Biggest Risk
With existing earnings and potential for continued volume growth	InnoLight, Skyetech (or names)	Overseas AI compute capital expenditure, 800G / 1.6T ramp-up, profit realization	Expectations are too high; even if financial reports continue to grow, they might pull back because marginal gains are no longer exceeding expectations.
Strong industry direction, but profit growth lags stock price movement	SMIC, Hygon Information (or names)	Domestic substitution, independent computing power, industrial security	Valuation runs first; it is easier to enter a sideways or consolidation period later.
Most aggressive earnings, but highest volatility	Cambricon (or names)	Simultaneous explosion of orders and profit, maximum elasticity	Once the market shifts from “looking at growth rate” to “looking at sustainability,” pullbacks are often the strongest.

So, if you’re asking about semiconductors, the chances are they should be fine/viable.

If you are asking whether adding positions at a spot like “2026-05-25” means catching a high price, then the answer should not be based on industry trends; rather, it should address position sizing and drawdowns:

For those who already have positions at low levels, the main question isn’t whether to chase or not, but whether to reduce the position exposure/scale back the size of their holdings.
For those with no positions at high levels who want to chase, what they are buying is mainly sentiment and expectations that emerged over the past six or seven weeks.
For those who genuinely want to get into the market, a more reasonable prerequisite is usually not a single large bullish candle, but rather a period of proper consolidation, or having the next earnings report push expectations further ahead.

Historically, what most resembles the present isn’t just a period like 2021 semiconductors or 2023 AI/CPO. The clustering of alcoholic beverages (Baijiu), new energy, pharmaceuticals, and stable blue chips are all reminding us of the same thing: the market is often most dangerous not when the underlying logic is weakest, but when the logic is smoothest and everyone can easily understand it.

I don’t think this current cycle in semiconductors is over, nor do I think it’s just a pure bubble. But if you translate “the industry fundamentals are sound” directly into “it’s fine to jump in today,” you are very likely to repeat the old pattern: the logic is real, and so are the drawdowns. What makes people miserable in the end isn’t realizing they picked the wrong direction; it’s looking back years later and finding out that what they bought wasn’t the industry trend, but rather the most expensive tail end of that consensus.

References

Yahoo Finance, historical quotes for 159995.SZ: https://finance.yahoo.com/quote/159995.SZ/history
Yahoo Finance, historical quotes for 515070.SS: https://finance.yahoo.com/quote/515070.SS/history
Yahoo Finance, historical quotes for 512690.SS: https://finance.yahoo.com/quote/512690.SS/history
Yahoo Finance, historical quotes for 515030.SS: https://finance.yahoo.com/quote/515030.SS/history
Yahoo Finance, historical quotes for 512010.SS: https://finance.yahoo.com/quote/512010.SS/history
Yahoo Finance, historical quotes for 510300.SS: https://finance.yahoo.com/quote/510300.SS/history
Yahoo Finance, historical quotes for 600519.SS: https://finance.yahoo.com/quote/600519.SS/history
Yahoo Finance, historical quotes for 000858.SZ: https://finance.yahoo.com/quote/000858.SZ/history
Yahoo Finance, historical quotes for 601012.SS: https://finance.yahoo.com/quote/601012.SS/history
Yahoo Finance, historical quotes for 300015.SZ: https://finance.yahoo.com/quote/300015.SZ/history
Yahoo Finance, historical quotes for 603288.SS: https://finance.yahoo.com/quote/603288.SS/history
Yahoo Finance, historical quotes for 688981.SS: https://finance.yahoo.com/quote/688981.SS/history
Yahoo Finance, historical quotes for 688041.SS: https://finance.yahoo.com/quote/688041.SS/history
Yahoo Finance, historical quotes for 688256.SS: https://finance.yahoo.com/quote/688256.SS/history
Yahoo Finance, historical quotes for 300308.SZ: <https://finance.yahoo.com/quote/300

Artificial Intelligence

Original Prompt

$blog-writer Did modules related to recent A-share surges, such as semiconductors and AI companies, experience similar trends before? When did this happen, what sectors were involved, what were the related stocks, how long did it last, and how did it collapse? Is increasing semiconductor holdings reasonable now, or is this buying at peak prices? Analyze the data of the related companies. All the data should be organized and compiled into tables for easy reading.

This piece establishes its core narrative, material density, and structure based on the original prompt provided above, following the initial drafting approach. The date field retains the original publication time, and all other content is dedicated solely to serving the scope of the current article.

Zhipu and MiniMax entering Hengke; rules and buying pressure are two completely different things.

Fri, 22 May 2026 20:04:43 +0800

After Zhipu and MiniMax were included in the Hang Seng Tech Index, the most common question that arises is: Since they have not yet reached their first lock-up period, are index funds compelled to buy shares?

This question cannot be answered directly based on emotion.

Inclusion in the index must first adhere to publicly disclosed methodologies. Delisting will affect future supply and stock price pressure, but it is not a hard threshold within the Hang Seng Tech Index methodology. Passive funds buy before and after the index takes effect because their goal is to track the index, not because the index company is arranging exits for existing shareholders.

I cannot see the “Unlisting” item in Hengke’s rules.

The methodology of the Hang Seng Tech Index is clearly written: it represents 30 companies listed in Hong Kong with high exposure to technology themes.

Several key conditions can be laid out first:

There is no “mandatory passing of the first restriction period” here.

So, if Zhipu and MiniMax meet the criteria for industry, theme, innovation screening, liquidity, and market capitalization ranking, they might enter Heng Ke (HK STAR Market segment). Whether they are one month away from the first layer lock-up period is another issue.

This is also where index rules and investment sentiment often clash. Investors care about whether current buying activity represents sustainable demand, while index methodology cares about whether it meets the sample selection criteria.

What Exactly Changed in This Announcement

The Hang Seng Index Company announced the quarterly review results on May 22, 2026. The announcement stated that all changes would be implemented after the close of trading on June 5, 2026, and would take effect from June 8, 2026.

The Hang Seng Tech Index maintains 30 constituent stocks, and the adjustments are:

Action	Company
Included	MiniMax Group Inc. - W `0100`
Included	Beijing Zhipu Huazhang Science & Technology Co., Ltd. `2513`
Excluded	Kingdee International Software Group Co., Ltd. `0268`
Excluded	Kingsoft Company Limited `3888`

Putting these four names together makes it easy to write “New AI players replacing old software companies.” However, this is merely a narrative construct. The actual rules are much colder: after quarterly reviews, the 30 positions will be re-ranked according to methodology, resulting in turnover.

What’s more worth looking at this time is not “who entered,” but how significant their weight/influence will be after joining.

Announcement Appendix Three lists the post-change weights according to the criteria of “assuming index adjustment on May 20, 2026.” The initial weights for Zhipu and MiniMax are not high:

Company	Circulating Coefficient	Adjusted Weight
Zhipu `2513`	8%	0.53%
MiniMax `0100`	6%	0.36%
Kingdee International `0268`	85%	Excluded before 0.70%
Kingsoft Software `3888`	75%	Excluded before 0.67%

The two newly added companies together account for less than 1%.

Therefore, it is true that “passive funds buy,” but “massive accumulation” cannot be simply inferred from merely being included [in an index]. The size of the buying force depends on the free float market capitalization, weight, tracking fund scale, and the degree of trading congestion within the effective window.

De-listing and Index Inclusion Are Not the Same Thing

The most critical point raised in the original question was: Since the shares haven’t been unlocked/released yet, will funds step in to buy support?

This needs to be broken down into three layers.

First, determine if the shareholder can sell.

If the lock-up period specified in the prospectus has not yet expired, the relevant shareholders are prohibited from selling on the open market. When index funds purchase shares, they can only buy them from the circulating supply in the secondary market and cannot directly convert locked-up shares into sellable stock.

Secondly, index rules—whether or not they are lifted/allowed.

The Hengke Methodology considers the sample space, technology themes, innovation screening, market capitalization ranking, quarterly reviews, free-float weight, and upper limit constraints. It does not include “the first lockup period has passed” as a screening criterion.

Thirdly, whether passive funds will buy.

They track Hang Seng Tech’s ETFs and index products, requiring adjustments to holdings around the effective date. They are not actively judging whether “this company is cheap or expensive,” but rather controlling tracking error.

When considering these three components together, the conclusion becomes much clearer:

This inclusion is not opening a selling channel for locked-up shareholders;
It will introduce predictable passive buying demand into the free float market cap;
If the free float is thin, price elasticity may be more pronounced;
However, this does not equate to “index funds covering the sales of newly unlocked shares.”

The more accurate statement is that while rules may disregard lock-up periods, capital focuses on the circulating supply (or free float).

Inclusion Does Not Guarantee Gains

Another common misconception: Does rising/entering an index automatically signal bullish sentiment, and does exiting/falling out of it necessarily signal bearish sentiment?

It may have short-term trading implications. Between the announcement and its effective date, active capital flows, arbitrage activities, ETF rebalancing, and derivative hedging will all move preemptively. The last closing auction before the effective date often makes it easier to observe mechanical buy and sell volume.

But in the medium to long term, things cannot simply be equated.

The Hang Seng Tech Index undergoes quarterly reviews. Often, by the time a company is included, its market capitalization, trading volume, and market hype have already completed a cycle; similarly, when it is excluded, its stock price and liquidity may have already undergone a period of weakness. Therefore, index adjustments are more like ex-post validation (or post-hoc confirmation), not the starting point for new fundamentals.

Therefore, when looking at index adjustments, I will divide them into three parts:

Question	What to Observe
Whether it can be included	Methodology, Industry, Theme, Innovation screening, Market cap ranking
Will there be passive buying demand?	Initial weight, Tracking fund size, Effective window
Will the price rise later?	Company fundamentals, Valuation, Liquidity, Market risk appetite

The first thing is the rule, the second thing is the transaction, and only then comes investment judgment.

If you mix them together, it’s easy to portray “index inclusion” as an infallible catalyst, or “not yet released/unlocked” as a perfect conspiracy.

My Perspective on Zhipu and MiniMax

ZhiPu and MiniMax’s inclusion in [HengKe context], suggests that Hong Kong-listed AI companies are beginning to enter the scope of index inclusion criteria. This matter holds symbolic significance as well as potential short-term trading implications.

But it is neither fundamental support nor a locked-up stock exit arrangement.

The three things that really need attention are:

First, whether the weights will continue to increase. Low initial weights do not mean they will remain low in the future. Changes in stock price, market capitalization, and float ratio will all affect rebalancing.

Second, changes in the circulating supply after restrictions are lifted. Index inclusion cannot resolve the fundamental supply problem. Once the lock-up period ends, the true circulating supply expands, requiring a reevaluation of price pressure.

Third, Hengke’s internal capital environment. The scale of tracking funds, ETF subscriptions/redemptions, and hedging using Hengke futures and options will all impact trading volatility before and after the effective date.

So I will not label this as a “buying opportunity” nor will I label it a “major upside.”

It is more like a validation of rules: AI new stocks meet the methodology and are included in the index; passive funds buy according to weightings; subsequent performance will continue to depend on fundamentals, circulating supply, and market sentiment.

One-sentence summary is:

The fund allocates investments according to weights, regardless of lock-up expiration rules. The index company is not responsible for any subsequent price increases or decreases.

References

Writing Notes

Original Prompts

$blog-writer Hang Seng Tech Index adjustments: Zhipu and Minimax are going to be included in the index. What are the rules for index adjustment? These two companies haven't reached their first lock-up period yet; by including them in the index, is it forcing funds to absorb/buy up shares? Please analyze the Hang Seng Tech Index, providing a list of changes from the last three years. Systematically review corresponding contracts: those that were included and those that were removed (kicked out), along with their fluctuation ranges over the next short period (one year) and their cumulative fluctuation range to date.

Writing Idea Summary

Retain core issues from the original prompt: rules, delisting, and passive fund buying pressure.
Remove large sections of historical change tables and price fluctuation tables from the old draft, to avoid information density overwhelming the main narrative.
Only retain key items concerning this round of adjustments, initial weights, and methodology; transform the article from a data repository into a rules explanation draft.

U.S. Treasury Yield at 4.5%: What Will I Actually Get After Buying?

Thu, 21 May 2026 10:16:49 +0800

When many people first see US Treasuries, the easiest thing to do is misread a number.

When you see 4.5% advertised in the market, it’s easy to mentally fill in a simple phrase: “If I invest now, I will reliably earn 4.5% every year and continue receiving it until maturity.”

This statement is only half right.

4.5% often aligns more closely with an “annualized yield metric,” and does not mean that you receive cash of 4.5% every year. What actually reaches your account is the result calculated from three factors combined: coupon payments, the purchase price, and principal repayment at maturity.

Conclusion First

If you purchase standard fixed-rate U.S. Treasury debt—meaning Treasury Notes or Treasury Bonds—and hold them until maturity without selling early, then:

What you receive is a rules-based cash flow, not an arbitrary percentage.
Interest is typically paid semi-annually.
The principal amount is returned at maturity.
The 4.5% seen upon purchase should be broadly understood as “the annualized yield expected near the time of maturity,” but it does not equal a flat 4.5% cash payout every year.

If you buy Treasury Bills, which are short-term treasury bonds maturing within one year, then it is an entirely different approach:

Usually, there are no interest payments during the term.
It is purchased at a discount and redeemed at face value upon maturity.
The profit mainly comes from “the purchase price being lower than the money received at maturity.”

Therefore, for newcomers to US Treasuries, remember this key point: First, distinguish between whether you are buying an interest-bearing bond (coupon bond) or a discount note.

What is the `4.5%` actually?

The three terms most often confused here are:

Term You See	What It Means	Does It Change?	What It Actually Affects
Coupon / Interest Rate	The rate set at the time of bond issuance	Fixed/Constant	Determines how much interest you receive each period
Yield to Maturity (YTM)	The annualized return calculated based on your current purchase price	Changes	Determines the yield scope for the investment from “now until maturity”
Price	The actual price you bought it at	Changes	Determines if you bought at a discount, par value, or premium

The U.S. Department of the Treasury states very clearly on TreasuryDirect: Notes and Bonds pay interest every six months, and the coupon rate is determined at auction; but bond prices fluctuate with yields. In other words, the yield is not the cash given directly to you; it is an annualized result calculated by combining the price and future cash flows.

This is also why common headlines in the news such as “10-year US Treasury yield rises to 4.5%” generally cannot be directly interpreted as “Buying 10-year US Treasuries now yields a 4.5% annual cash dividend.” This metric is more accurately described as the yield convention reported by the market for this type of maturity bond. My explanation here is based on TreasuryDirect’s definition regarding the relationship between coupon, price, and yield.

Calculating with an Official New Bond

Mere discussions of concepts are useless; one must speak based on the official auction results.

The U.S. Treasury announced the auction results for a 10-Year Note on 2026-05-12:

Interest Rate: 4-3/8%, which is 4.375%
High Yield: 4.468%
Price: 99.256552
Issue Date: 2026-05-15
Maturity Date: 2036-05-15

If you purchase based on this auction result with a $1,000 face value, the calculation would be as follows:

Item	Value	Explanation
Face Value	$1,000	Basis of the principal recovered at maturity
Purchase Price	$992.57	Because the price was `99.256552`, you

The dividend thing, the formula is actually simple:

\[ \text{Half-year coupon} = \text{Face Value} \times \text{Coupon Rate} \div 2 \]

Substituting this debt, it is:

\[ 1000\times 4.375\% \div 2 = 21.875\text{ USD} \]

If you do not sell over these 10 years, and excluding taxes and exchange rates, the nominal total cash flow would be:

\[ \text{Total Cash Flow}=\text{Total Coupon}+\text{Maturity Principal}=437.50+1000=1437.50\text{ USD} \]

Your initial purchase cost was approximately $992.57. Therefore, from the perspective of “total amount recovered,” this debt is clearly not a loss.

However, let’s pause here. This does not equal you receiving 4.468% in cash annually.

What you actually receive is:

Dividend credited semi-annually.
The principal is recovered on the final maturity date.
Because you purchased below face value, you will also gain an additional spread from “returning from the purchase price to the face value.”

When combined, these three parts make up the 4.468% High Yield shown in the auction results.

Why is the coupon rate 4.375%, but the yield to maturity is 4.468%?

Because you bought it below face value.

The coupon rate for this bond is only 4.375%, but the transaction price is 99.256552, which is below its face value of 100. Therefore, although you receive annual interest calculated on the full face value (4.375%), you will still be repaid the principal of 100 at maturity. This small discount built into the price will lift the overall holding-to-maturity yield to 4.468%.

The converse is also true.

If a bond’s coupon rate is higher than the prevailing market yield, it tends to trade at a premium. This means you have to pay more upfront when you buy it. Consequently, although you receive a larger coupon payment every six months, upon maturity, you only get back the face value. The excess paid (the premium) will slowly erode over time, ultimately pulling the overall yield down closer to market levels.

Therefore, when looking at US Treasuries, the most reliable sequence is not to look at 4.5% first, but rather:

First, check what type of bond it is and its remaining maturity period.
Next, look at the coupon rate.
Finally, examine the yield corresponding to your actual purchase price.

Whether the Yield Can Truly Be Understood as `4.5%` if It Never Sells

Yes, but two footnotes are needed.

First, `4.5%` is more like an annualized rate, not a cash payout rhythm/disbursement schedule

Provided that you buy a fixed-rate Note or Bond, and actually hold it until maturity without selling it midway, what you lock in at the moment of purchase is a stream of future cash flows:

Interest paid every six months
Principal repaid at maturity
How much do you pay now for this stream of cash flows?

The market-provided Yield to Maturity is essentially calculating these three factors into an annualized rate of return. For beginners, you can understand it as: “If this bond is held until maturity, its approximate yield level will be this.”

Secondly, the money in the account will not automatically compound `4.5%`

This point is also easily misunderstood.

The YTM of a bond uses an annualized yield basis, but if the coupon payments you receive halfway are just sitting idle in your account without being reinvested, they will not generate the same rate of return on their own.

In other words:

The inherent cash flow of the bond, which you can generally estimate in advance.
Your final compounded result, which depends on how you handle the coupon payments after receiving them.

A more precise way to say this is: the 4.5% shown upon purchase is more like a “reference for annualized yield until expiry,” rather than “a guarantee of 4.5% automatic annual compounding in the account.”

When Will It Be Issued/Distributed

Approach this issue by focusing on different segments/areas; don’t try to solve everything all at once.

(Alternative translations depending on context:)

Don’t use a blanket approach; tackle it by type. (More formal)
Segment this matter and don’t try to grab everything at once. (Closer to the literal meaning but still clear)

`Treasury Bills`

The official stance of the U.S. Department of the Treasury is that Bills are treasury securities with a maturity of one year or less, typically issued at a discount and paid at face value upon maturity.

For you, it is:

Requires an upfront investment upon purchase
Usually no regular coupon payments during the term
Face value is returned in a lump sum at maturity

So, if you buy short-term bonds, when asking “how often is the interest paid?”, the answer is often: It’s neither paid monthly nor semi-annually; there are no payments in between—it all settles at maturity.

`Treasury Notes` and `Treasury Bonds`

These two types are coupon bonds (or interest-bearing bonds). The US Treasury Department clearly states: a fixed rate, with interest paid every six months until maturity.

To you, it is:

Receive coupons semi-annually
Recover principal at maturity

There is also a detail that can easily confuse beginners: The U.S. Treasury mentioned in its statement noted that when purchasing during certain ‘reopening’ periods, the price might include a small amount of accrued interest; this portion will be returned/credited out during the first formal dividend payout. In other words, the initial amount you receive may occasionally differ from what your intuition suggests, and it doesn’t necessarily mean the system calculated it incorrectly.

I think the most important thing to clarify first is not the yield rate, but the definitions/scope

If this is your first time buying US Treasuries, I actually wouldn’t recommend immediately fixating on whether “4.5%” is high or low.

Understanding these next three points will automatically help you avoid half of the potential pitfalls that follow:

Yield is not equal to Dividend.
Holding-to-Maturity Yield is not equal to Annual cash distribution ratio.
Never selling can only help you avoid the impact of intermediate price fluctuations on the sale result, but it cannot avoid exchange rates, taxes, and brokerage fees.

In this article, I deliberately did not elaborate on exchange rates, taxes, or brokerage fees. It’s not that they aren’t important, but if these three variables are introduced together, the inherent yield mechanism of US Treasuries can become easily complicated/confused. If you first clarify how “bonds themselves pay money,” then tackle the layer of cross-border investment, it will be much easier to grasp.

References

TreasuryDirect, Treasury Notes
TreasuryDirect, Understanding Pricing and Interest Rates
TreasuryDirect, Buying a Treasury Marketable Security
U.S. Treasury, Treasury Auction Results, 10-Year Note, May 12, 2026

Writing Notes

Original Prompt

Detailed explanation of US Treasury yields. For example, if a bond yield is currently priced at 4.5%, what kind of return can I expect if I buy it now and hold it indefinitely? When does it pay out? I am a complete novice when it comes to US Treasuries.

Writing Outline Summary

First, separate the concepts of coupon, yield, and transaction price. Otherwise, the subsequent analysis of cash flow will be incorrect.
Use the official 10-year US Treasury auction results from 2026-05-12 as an example, rather than just providing abstract definitions.
The main body should focus on answering “what money you will receive and when,” instead of giving a general discussion of macroeconomic interest rates.
Deliberately separate the writing for Bills and Notes/Bonds

Extended Brainstorming

Topic	Inclusion Status in Main Body	Reason
Use official 10-year auction results as an example	Include	Has specific numbers, which can clarify the difference between a `4.375%` coupon rate and a `4.468%` yield.
Distinguish between `Bills` and `Notes/Bonds`	Include	The user asks "

Why didn't the Hang Seng Tech index hit new highs alongside A-shares?

Wed, 20 May 2026 20:38:09 +0800

Bottom line first.

As of May 20, 2026, the strength of A-shares and the weakness of Hang Seng Tech are not due to one asset pool having lagging components; rather, it is a divergence between two sets of pricing logics. On May 20, 2026, the Shanghai Composite Index closed at 4162.19 points, still fluctuating near 4200; meanwhile, on its most recently available closing date of May 19, 2026, the Hang Seng Tech Index closed at 4857.46 points, which is still short of the phase historical high point of 10945.22 recorded on February 17, 2021, by 55.6%.

If your statement that the “CSI 300 ETF has matched its historical peak” refers to the most common 510300, the closing price I captured on 2026-05-20 was 4.871. This still represents about a 16.1% drawdown from the 5.807 recorded on 2021-02-10, and has not reached an all-time high. It is highly likely that various calculation bases were mixed here: Price, Net Asset Value (NAV), Adjusted NAV, Total Return Index. They look like they are the same thing, but they actually are not.

Allow me to correct a name. The “China Golden Dragon Fish Index” you mentioned usually refers to the Nasdaq Gold Dragon China Index, whose English name is Nasdaq Golden Dragon China Index, not “Golden Dragon Fish.”

Clarifying the Scope

This time, I will only calculate the correlation using rates of return, rather than calculating it directly using price levels. Price levels have long-term trends; if we calculate the price correlation directly, it is very easy to mistakenly consider any two assets that have risen over the long term as being “strongly correlated.”

Object	Definition Used	Characterized As
Hang Seng Tech Index	30 large technology companies listed in Hong Kong, defined by the Hang Seng Index Company, with a single weight cap of 8%	HK tech leaders + offshore China growth
Nasdaq Golden Dragon China Index	A collection of securities defined by Nasdaq that are listed, registered, or primarily linked to mainland China.	US-listed Chinese concept stock risk appetite
Nasdaq Composite Index	Large-cap US tech/growth factor in the American stock market.	Global USD technology asset risk appetite

Hang Seng Tech and A-share broad market indices both appear to be called “China assets,” but their components are vastly different. The top ten weights listed in the official Hang Seng Tech factsheet for April 2026 basically include Meituan, SMIC, BYD, Alibaba, NetEase, Xiaomi, Tencent, JD, Baidu, and Kuaishou. It is not focused on banking, insurance, resources, or dividends, nor is it an index that “will definitely rise just because China assets rise.”

Why A-Shares Are Strong While Hang Seng Tech Is Pulling Back

I am more inclined to break this down into three layers.

First layer, compositional structures are different.

The recent strengthening of the SSE Composite Index and CSI 300 tends to benefit finance, core manufacturing, large-cap weights, and certain high-dividend assets more significantly. Hang Seng Tech operates on a different playbook: internet platforms, semiconductors, consumer electronics, and smart vehicle supply chains. These companies are more sensitive to valuation, overseas liquidity, risk appetite, and geopolitical sentiment.

Second layer, the source of funds for pricing differs.

A-shares are more a reflection of local capital, local policies, and local risk appetite. Although Hang Seng Tech is also influenced by southbound funds, it is fundamentally an offshore market asset. The offshore market assigns higher weight to “regulatory expectations, US dollar interest rates, ADR risk premium, and overseas fund positioning.” When A-shares pursue a trend of “stable growth + large market weight,” Hang Seng Tech may completely undergo a cycle of having risen too quickly previously and now retracing some gains.

Third layer, that previous uptrend was fundamentally not on the same beat/pace.

Over the past two years, Hang Seng Tech has experienced several very steep rallies; its uptrends tend to be faster than A-shares, but its pullbacks are also more severe. According to the 1-year annualized volatility provided in the Hang Seng Index Company’s April 2026 factsheet, Hang Seng Tech stands at 30.80%, while HSI is only 14.08%, and the SOE index is 12.62%. High-volatility assets are more prone to premature profit realization during periods of divergence.

I would summarize this phenomenon in one sentence: This round of A-shares is more like “local blue-chip asset recovery,” while Hang Seng Tech is more like the “re-pricing of offshore Chinese growth stocks.” The two will move in the same direction, but they will not be synchronous.

Are Hang Seng Tech and Chinese Concept Stocks Strongly Correlated?

Conclusion first: Yes, very strong at week and month scales; medium-to-high correlation at daily scale.

I aligned the common trading dates from 2020-08-17 to 2026-05-19, and calculated Pearson correlation using daily return, 5-day return, 20-day return, and 60-day return. The results are very straightforward:

Pair	1-Day Return Correlation	5-Day Return Correlation	20-Day Return Correlation	60-Day Return Correlation
Hang Seng Tech vs Nasdaq China Dragon	0.60	0.85	0.90	0.93
Nasdaq China Dragon vs Nasdaq Composite	0.49	0.46	0.35	0.35
Hang Seng Tech vs Nasdaq Composite	0.17	0.36	0.31	0.28

This result is very interesting.

Hang Seng Tech and China Concept Dragon are, in the short to medium term, two exchange listings that can be viewed as essentially the same type of risky asset.
China Concept Dragon and the NASDAQ are not as closely linked as people imagine. They are correlated, but not bound.
Hang Seng Tech and the NASDAQ are weaker/less coupled. They are separated by a layer of “China assets’ own risk premium.”

Another detail worth noting is…

Many people think that when Hong Kong stocks open during the day, they should primarily follow the movements of the US market from the previous night. However, if we truly align the daily rise and fall of Hang Seng Tech with “the China Concept Golden Dragon” from the previous trading day, the correlation actually diminishes. This indicates that it is not simply “blindly following” last night’s US market, but rather being driven by the same round of offshore China beta along with the Chinese concept golden dragon.

China Concept Gold Dragon and NASDAQ: Is there a strong correlation?

My assessment is: Moderate correlation, not strong correlation.

Simply noting that they are all traded in the U.S. can lead to a misunderstanding. Although the China Concept Gold Dragon Index is listed on US exchanges, its definition tracks Mainland Chinese companies, not domestic American tech leaders. Its movements are influenced by two categories of factors:

One type concerns USD growth stock risk preferences represented by the NASDAQ.
Another type relates to China’s policy, audit regulation, ADR discount, RMB expectations, platform economy, and real estate sector risks.

This explains why the daily return correlation for the China concept gold dragon and Nasdaq was only around 0.22 in 2024, but returned to around 0.59 by 2026. It is not a permanently stable US tech beta; it will periodically revert/switch back to a “China asset beta.”

Is There a Cycle Where US Stocks Drop and Hang Seng Tech Surges?

Yes, and more than once.

However, to provide the full context: It was not that Hang Seng Tech decoupled from US stocks; rather, during that period, the force of Chinese asset revaluation outweighed the Nasdaq adjustment.

I’ll select three representative scenarios:

Observation Window	Hang Seng Tech	Nasdaq China Dragon	Nasdaq Composite	Explanation
60 trading days as of 2023-01-19	+58.8%	+72.0%	-0.9%	Policy recovery after extreme undervaluation in 2022 + reopening trades
60 trading days as of 2024-10-07	+49.6%	+35.4%	-3.9%	Cyclical revaluation of Chinese assets, with greater elasticity for Hang Seng Tech
120 trading days as of 2025-03-18	+67.2%	+42.8%	-2.8%	HK tech and Chinese concept stocks experiencing a strong, independent recovery phase

Therefore, the answer is not “no,” but rather “yes, although it rarely occurs during periods of pure global risk sentiment.” When Hang Seng Tech manages to move counter-cyclically against US stock pullbacks, the underlying causes are usually more localized catalysts such as marginal changes in Chinese policy, valuation recovery, and replenishing offshore Chinese capital positions.

How Did Each Major Cycle in Hang Seng Tech Start and End?

I did not count every 10% rebound as a new cycle, as doing so would turn the article into a market report/broadcast. Instead, I used a swing-based segmentation method with a threshold close to 30% to divide the period from 2020-08-17 to 2026-05-19 into major levels, and then marked significant minor drawdowns separately.

In this table/chart, what I value most is not the magnitude of fluctuation, but its manner of conclusion (or how it ends).

The upward trends in the Hang Seng Tech sector rarely proceed gradually; rather, they tend to be:

Valuation preceded by a long preparatory period;
Sudden consolidation of sentiment;
Finally ended with a sharp sell-off.

The downward movement will not be a simple linear drop to rock bottom; instead, it is:

The main decline period is very long;
Interspersed with several decent rapid rebounds;
When truly bottoming out, it is usually accompanied by “marginal policy changes + very light positions + low valuation.”

The most questionable aspect of these assets is here: It looks like an index, but it’s actually more like a highly volatile basket of sectors.

The Last Sentence

If you treat Hang Seng Tech as a “Hong Kong Nasdaq,” it’s easy to misjudge; if you see it as the “China tech growth beta” within Hong Kong stocks, many things become clearer.

A-shares being strong does not mean that the Hang Seng Tech must be equally strong. The Hang Seng Tech is strongly correlated with China’s concept stocks, but only moderately correlated with the Nasdaq. When US stocks fall, the Hang Seng Tech can indeed rise sharply, but that often indicates that the dominant factor has switched from “global tech risk appetite” to “China’s assets being re-evaluated on their own merit.”

So for this market trend, don’t just focus on one direction. First, see who it is following.

References

Hang Seng Indexes, Hang Seng TECH Index Factsheet, April 2026
Nasdaq, Nasdaq Golden Dragon China Index Methodology
FRED, NASDAQ Golden Dragon China Index
FRED, NASDAQ Composite Index
CSI Index Company Limited, CSI 300 Factsheet, 2026-03-31
Sina Finance, Hang Seng Tech Index Page

Writing Notes

Original Prompt

The Shanghai Composite Index for A-shares has risen significantly recently but also corrected once, fluctuating around 4200. However, the CSI 300 ETF is approaching historical peak levels. Investigate why the Hang Seng Tech Index declined during the same period. Is the Hang Seng Tech Index highly correlated with the US China Dragon Fish Index? Is the China Dragon Fish Index highly correlated with the Nasdaq Index? Are there cycles where the Hang Seng Tech Index surges when US stocks fall? Analyze the duration of previous cycles, and how they started and ended.

Writing Idea Summary

Correct “Golden Dragon Fish Index” to Nasdaq China Dragon Index, and separate the metrics for ETF price, adjusted net value, and total return index.
Do not calculate correlation directly based on specific levels; only use rates of return, to avoid long-term trends artificially inflating the correlation.
First address “why there is a desynchronization,” then answer “whether it is strongly correlated,” and finally focus on the major cycle of Hang Seng Tech.
Anchor relative time to specific dates (2026-05-19 and 2026-05-20) for terms like “recent” or “same period,” to prevent misleading interpretations based on relative timing.
Deliberately avoided elaborating on individual stock financial reports and single-day news headlines, because the more important focus of this article is the index structure and capital pricing logic.

Extended Brainstorming

Direction	Whether to include in body	Handling Method
Day-by-day alignment of Hang Seng Tech and Southbound Net Inflows	No	Valuable, but it changes the article from an index comparison to a capital flow monitoring report, risking topic drift.
Decomposing A-shares into SSE 50, Dividend Stocks, CSI A500	No	Can explain in greater detail, but will dilute the main line of argument (“Why Hang Seng Tech is unsynchronized”).
Recalculating “whether it'

Relying solely on stock prices to gauge the semiconductor cycle is insufficient; SK Hynix's earnings report serves as a better barometer.

Wed, 13 May 2026 21:25:24 +0800

In the previous article The end point of this semiconductor cycle is probably not in 2026, I presented my conclusion first, but deliberately did not dive too deep into the specific details of the financial reports.

What we are covering this time is the part that is most easily obscured by market sentiment: When semiconductors rise, everyone knows they are profitable; but what truly determines whether a cycle can be extended or which company can capitalize on high growth more thoroughly is often not the stock price, but rather the profit and loss statement, capital expenditure, and product investment direction during the trough.

If I must make a more specific judgment, as of May 13, 2026, I still do not pinpoint 2026 as the end of this upcycle. However, if I have to pick just one major player among the giants that is most worth watching, it would be SK hynix. Not because it hasn’t gone through a downturn—quite the opposite—but because it made the most representative strategic choices when things looked their worst in 2023.

First, standardize the metrics/scope; otherwise, the table may be misleading

This article only selects three companies: SK hynix، Micron، TSMC.

The reason is simple:

SK hynix: The most direct beneficiary of AI memory this round, and also the data object that needs special focus in this article.
Micron: The purest storage beta among US stocks, useful for comparison with SK Hynix.
TSMC: It is not a memory company, but precisely because of that, it is suitable for use as a control group.

To clarify, there are two approaches/perspectives to explain first:

SK hynix, TSMC are primarily based on calendar years.
Micron is based on fiscal years, with FY2025 as of August 28, 2025.

Therefore, these tables are better suited for analyzing period-over-period changes within a single company, rather than being used to perform forced cross-sectional valuations across different companies.

With two downcycles and a bounce/rebound, memory’s reported earnings resilience is striking

After reading this table, the conclusion on the first layer is actually already out:

Semiconductors are not a cycle; at least, memory and logic foundry work are not.

In 2019, the global semiconductor industry was already in a downturn, but TSMC’s revenue still managed to grow against the trend, and its profit margins did not collapse. Conversely, when SK hynix and Micron encountered a storage price cycle reversal, their operating profits would decline before (or more steeply than) their revenues, and the magnitude of this decline is much greater.

This is also why I kept emphasizing in the previous article that US memory and Korean semiconductors cannot be viewed merely as “ordinary semiconductor stocks.” They are more aggressive during upcycles, but they are also harsher during pullbacks.

What really needs to be addressed is how Hynix’s “counter-cyclical” nature actually works.

If you only look at the results, SK Hynix’s financial reports for 2024 and 2025 truly seem like they are “printing money.”

Year	Revenue	Operating Profit	Operating Profit Margin
2023	32.766 trillion KRW	-7.73 trillion KRW	-23.6%
2024	66.193 trillion KRW	23.4673 trillion KRW	35.5%
2025	97.1467 trillion KRW	47.2063 trillion KRW	48.6%

From 2023 to 2025, this is not a normal recovery; it is a turnaround from deep losses to an operating profit margin approaching 50%.

But if “counter-cyclical expansion” is phrased too roughly, it will lose the key point.

A more accurate way to say it is:

\[ \text{Hynix's Counter-Cycle} \neq \text{Massive Expansion Across All Lines} \]\[ \text{SK Hynix's inverse cycle} = \text{Total investment contraction} + \text{HBM/DDR5/LPDDR5 momentum} \]

In other words, instead of heavily supporting all product lines equally in 2023, they prioritized reserving funds for the few lines expected to be most profitable in the next cycle during the worst times.

The difference is very big. (Alternatively: The difference is huge./There is a significant difference.)

The statements made in 2023 basically revealed Hynix’s strategy.

I’ll organize the official narrative chronologically; that way, it will be much clearer.

Therefore, if you remember the incident of “Hynix’s counter-cyclical capacity expansion,” that direction of memory is correct, but a qualifier must be added: it was selective counter-cyclical expansion, not comprehensive and risky overexpansion.

This is also why I feel it is the most accurate indicator.

Because what truly determines if this cycle of memory can be extended is not whether “everyone expanded production,” but rather who managed to secure critical market positions for HBM, DDR5, advanced packaging, and high-value server storage during the downturn.

Comparing it alongside Micron and TSMC makes SK Hynix’s characteristics even more obvious

Micron has actually done similar things before.

In FY 2023, its revenue dropped from $307.58 billion to $155.40 billion, and operating profit fell from $97.02 billion to -$57.45 billion; however, by FY 2024, revenue recovered to $251.11 billion, and operating profit turned positive at $13.04 billion. In FY 2025, it surged further to $373.78 billion in revenue and $97.70 billion in operating profit, with annual capital expenditures also rising from $81.2 billion in

What does this mean?

Micron is also preparing for the next phase of AI memory demand, and capital expenditures have clearly begun to turn upwards again.

However, Hynix’s characteristic remains more pronounced. This is because even when it was at the deepest trough in 2023, it continuously included HBM and DDR5 in its investment priorities multiple times. While Micron seems to be increasing investments after confirming a recovery, following demand trends; Hynix appears to have preemptively captured the next major growth trajectory even during the most challenging reporting period.

Looking at TSMC again, it feels completely different.

In 2019, the global semiconductor industry was already slowing down, yet TSMC raised its capital expenditure to $14.9 billion in 2019, citing stronger demand for 7nm. By 2023, even though the industry was digesting inventory, its revenue only declined by 4.5%, and operating profit decreased by only 17.8%. In 2024, it quickly rose back to NT$2.8943 trillion in revenue and NT$1.3221 trillion in operating profit.

This precisely illustrates a counter-intuitive but very important fact:

Under the term “semiconductor giants” lie two completely different financial structures.

The leader in memory is more like a high-leverage cyclical asset: when prices rise smoothly, profits explode; but when prices reverse, profits decline sharply.
The leader in logic / foundry is more like an industry platform with a moat. While it is also cyclical, unlike memory, it is not as easily overthrown by ASP.

So, if you want to focus on “companies best positioned to amplify cyclical upturns” this round, Hynix and Micron are certainly more sensitive than TSMC.

But if you want to monitor “when risk shifts from selective capacity expansion to industry-wide capacity expansion,” then TSMC’s massive capital expenditures, advanced packaging, and customer structure are instead another type of forward indicator.

These financial figures perfectly complete the analysis from the previous section

In my previous article, I said that this semiconductor cycle is unlikely to end by 2026. (or: …is highly unlikely to collapse before 2026.)

After supplementing the financial data in this document/report, the logic becomes much more coherent/complete.

First, the trough in 2023 was too deep.

Both SK hynix and Micron have experienced significant dips in their profitability/profit statements. The deeper this trough, once prices and product structures rebound, profit recovery is more likely to appear “jump-like” rather than “linear.”

Secondly, companies like SK Hynix have proven that establishing a strong market cycle was not something they began preparing in 2024, but rather positioning themselves since the toughest period of 2023.

This means that the high profits projected for 2026 are not merely residual from inventory rebuilding; rather, they represent the realization phase of the preceding round of selective counter-cyclical investments. As long as this realization period continues, the overall cycle is unlikely to end quickly.

Third, what is truly dangerous now is not Hynix securing the HBM line in 2023, but whether more and more companies will convert this selective investment into large-scale investment between 2025 and 2026.

Once the market shifts from focusing only on “high-end memory expansion” to general upward expansion by everyone, then the dangerous window I mentioned in my previous article will draw ever closer.

If we are going to follow up, I will only focus on these few matters

Metric	Current Implication	What it suggests if things start to sour/change
SK hynix / Micron’s capex guidance	Still boosting investment in AI memory, indicating the boom cycle is not over.	If capex expansion also targets standard DRAM / NAND, the seeds of oversupply risk begin to sow.
Share of HBM in DRAM revenue	High-value-added products are still driving up the profit

It is said that many people monitor the semiconductor cycle, developing the habit of looking at stock prices first, then checking the news, and only finally reviewing the financial reports.

However, if you truly want to pinpoint the time window this time, it is best to reverse the order.

First, look at how profits and product structures have changed according to the financial report; then look at where capex is being invested; finally, only then check whether the stock price has fully factored in/priced in two or three years of good growth.

SK hynix deserves to be singled out not because it is immune to cyclical downturns, but because during the most challenging memory market cycle of 2023, it provided a very standard answer:

We can scale back overall investments, but we cannot stop pursuing the few lines/trends that will generate the most profit in the next round.

This sentence might be more useful than many grand narratives of “semiconductor boom continuation.”

References

Writing Notes

Original Prompts

The previous articles suggested that the conclusion of this semiconductor cycle is unlikely to be in 2026. Since detailed financial data is lacking, I will write a new article supplementing the financial records of several major semiconductor players across multiple cycles. Specifically, Hynix—I recall news reports about Hynix expanding capacity during counter-cycles.

Writing Approach Summary

This piece does not repeat the cycle conclusions from the previous article; it only supplements details on financial reports, profit margins, and capital expenditure.
The main text intentionally places memory and logic/foundry side-by-side, but the focus is on observing changes within a single company across different cycles, rather than making forced comparisons.
Hynix’s “counter-cycle” status has been refined into more accurate descriptions: total investment shrinks, but HBM and DDR5 development remains continuous.
This article intentionally avoids dedicating a large section solely to Samsung, as that would pull the focus away from “why Hynix acts like a barometric indicator” back to broader comparisons of Korean semiconductor companies.
Spot price curves and equipment chains are also worth discussing, but we will omit them in this piece to prevent the single article from becoming too scattered.

Expanded Brainstorming

Direction	Inclusion Status in Main Body	Reason
Samsung single-column years financial data	No	Worth adding, but will stretch the scope into a comparison of two Korean giants, weakening the focus on Hynix.
Spot price, contract price, inventory days curve	No	Helpful for judgment, but requires more precise criteria and continuous updates; easy to lose focus in a single article.
Why TSMC survived 2019	Partially Included	Needs a control group to prove that “semiconductor leader” does not have the same financial flexibility.
Whether Micron’s revised upward estimate of capex will be a trap	Partially Included	This is an extension line from the previous article’s judgment, and viewing it with Hynix has more comparative value.
Expansion of local Chinese storage and mature processes	No	Important for the supply side, but if expanded in this article, the main focus will shift from Hynix to a full industry supply map.

The endpoint of this semiconductor cycle is unlikely to be in 2026.

Tue, 12 May 2026 00:19:22 +0800

Regarding this round of semiconductor trends, I temporarily do not see a peak in 2026.

If forced to give an initial judgment, as of May 12, 2026, I am more inclined to place the truly critical period between the second half of 2027 and the first half of 2028, rather than now. The core driver of this current uptrend—particularly in US listed storage and Korean semiconductors—is not a general recovery, but rather AI pulling HBM, DDR5, and enterprise SSD up simultaneously. If supply expansion fails, both prices and profits will rise together.

This also explains why companies like Micron, SK hynix, and Samsung seem to be “printing money” lately. The semiconductor cycle hasn’t vanished, but this time it is unlikely to collapse when demand first kicks in; rather, it is more likely to crash when capacity expansion finally catches up, and the market has already front-loaded two or three years’ worth of profit.

Executive Summary: Why I Doubt the ‘End by 2026’ Prediction

Place a few pieces of primary and semi-primary source materials on the table first.

Connecting these signals makes the answer clearer.

This cycle is neither the traditional PC cycle, nor is it driven by smartphone replacement cycles, and certainly not a simple rehash of the “semiconductor shortage due to the pandemic.” Its engine is the capital expenditure from major hyperscale AI data centers, and the massive throughput requirements for memory and storage needed by AI servers are simultaneously stressing HBM, DDR5, high-capacity DIMMs, and enterprise SSDs. Micron was quite direct in its investor materials last December: “Overall industry supply is significantly below demand in the foreseeable future, and this tightness will persist well beyond 2026.”

In other words, the market isn’t betting on “whether things will be better next year,” but rather on “whether the period of tension before 2027 will end.” If this premise is not broken, the cycle will struggle to wind down on its own by 2026.

Why Did This Round Ignite Both US Storage Stocks and Korean Semiconductors

Many people tend to view the US memory/storage sector and South Korean semiconductors as two separate issues. In reality, these two trends are expected to largely converge by 2026.

Micron represents the most direct memory sector beta on US stock markets. Reuters, in a report on March 19, 2026, noted that Micron’s stock price has already risen by over 61% this year. Furthermore, the company increased its capital expenditure plan for fiscal year 2026 by another $5 billion, bringing the total amount to over $25 billion. This move itself indicates one thing: management acknowledges that existing capacity is insufficient to meet the demand generated by the current wave of AI development.

Korea was even more dramatic. SK Hynix hit an all-time high on May 4, 2026, closing up 12.52% that day. The same Reuters report also mentioned that high-ranking officials from the Bank of Korea judged that this chip cycle upturn might last longer than previous cycles. This statement is not groundless, as the two most core memory manufacturers in Korea have repeatedly released similar information during their earnings reports and conference calls.

Samsung even stated directly during its conference call at the end of April 2026 that, “based solely on the already received demand for 2027, the supply-demand gap in 2027 will be even larger than in 2026.” This statement is very significant. It means that the market’s current surge isn’t due to people competing for 2026 earnings, but rather they are preemptively trading the 2027 shortage.

Therefore, the most crucial characteristic of this market cycle is not the “overall semiconductor boom,” but rather:

AI has turned the most lucrative memory sub-sector into a bottleneck for the entire industrial chain.
HBM will consume DRAM capacity and advanced packaging resources, simultaneously constraining commodity DRAM as well.
The demand for SSDs in servers and the inference side no longer simply follows PCs or mobile phones; rather, it is directly tied to large model infrastructure.

When these three factors are combined, Micron, SK hynix, and Samsung’s profit elasticity will be enormous. The memory industry is inherently a high operating leverage sector, so when ASP rises, profits often do not increase linearly; instead, they jump significantly.

How Have Past Semiconductor Cycles Typically Concluded?

I don’t want to turn this section into a semiconductor chronicle (or history). What is truly powerful/disruptive in terms of investment, and therefore worth remembering, are actually the following cycles.

Cycle Phase	Approximate Duration	Termination Details	Data Observed at End of Cycle
1998-2000 Uptrend	Approx. 2 years	Dot-com bubble burst, decline in PC and mobile demand, overstocking pressure	Global semiconductor sales dropped from $2.04 trillion in 2000 to $1.39 trillion in 2001, a year-on-year drop of 32%
2002-2007 Uptrend	Approx. 6 years	Global Financial Crisis first depressed valuations, then suppressed end-user demand	Sales dropped from $255.6 billion in 2008 to $248.6 billion, and again by 9% to $226.3 billion in 2009
2016-2018 Uptrend	Approx. 3 years	Slowing growth rate in the second half of 2018, price cycle reversal combined with trade friction	Global sales dropped to $412.1 billion in 2019 (a year-on-year drop of 12.1%); memory sales dropped 32.6%, and DRAM dropped 37.1

Looking at this table, the pattern is actually quite straightforward.

The semiconductor cycle truly ends not usually because of “overvaluation, so it must fall.” It often requires two out of three conditions (or even all three) to appear simultaneously:

\[ \text{Cycle Peak} \approx \text{Supply growth rate catches up to demand growth rate} + \text{Inventory reversal from low levels} + \text{Marginal weakening of end-user demand} \]

To put it more simply:

The release of new capacity begins.
Clients are shifting from frantic purchasing to cautious observation.
Leading companies no longer talk about shortages but start discussing inventory, cost, and pricing pressures.

Died in a demand collapse and inventory in 2001.

Died in the macro crisis of 2008–2009.

Died in the price cycle and trade disruption of 2019.

De-stocking following the depletion caused by the pandemic in 2023.

So, one theory I am most skeptical of right now is that “the semiconductor has risen too much, so it must end soon.” Cycle peaks do not form this way. They require the underlying supply and demand relationship in reality to ease first.

What exactly is the difference between this round and the 2021-2022 round?

During the 2021-2022 cycle, many chips saw price increases, but the underlying logic was rather scattered. Automotive, home appliances, mobile phones, PCs, and servers were almost all engaged in restocking inventory, while the pandemic severely disrupted the supply chain. The conclusion of that cycle was also quite typical: end-user demand declined, channel inventories remained high, and the industry began a collective correction.

The cycle in 2026 will be more concentrated and more dangerous.

The focus is on money being poured into AI data centers. Although there are fewer sources of demand, the intensity at specific points is much greater.

The danger is that these types of needs are not everyday consumer goods in a completely market-driven way, but rather driven by the capital expenditure of several super large companies. Once cloud vendors and platform providers discover:

GPU utilization was not as high as expected;
Inference revenue realization is slower than capital expenditure;
Agentic AI has failed to achieve commercialization;
Or macroeconomic environment, tariffs, energy, and exchange rates are extending the return cycle.

The demand growth rate will drop very quickly.

The industry has not entered this stage yet. On the contrary, what can be seen is that multiple tech giants are continuing to heavily invest in AI infrastructure. Under SIA figures, global semiconductor sales already reached $791.7 billion in 2025 and are projected to approach one trillion US dollars by 2026. Samsung also clearly stated that server memory demand will remain strong in the second half of 2026 as hyperscalers accommodate enterprise AI and LLM services.

This is why I judge that: 2026 looks more like a year of concurrent price increases and expanded production, rather than a peak year.

When will this round most likely end?

I’ll give you a baseline assessment; nothing too mystical/over-the-top. (Alternative translations depending on context: I’ll provide a basic evaluation, keep it simple.)

My Benchmark Scenarios

If there is no sudden global recession, nor a dramatic collapse in AI capital expenditure, this semiconductor uptrend is more likely to enter a danger zone starting in the second half of 2027, and genuine cyclical peak characteristics will be easier to observe in the first half of 2028.

There are four reasons.

First, supply has a physical lag.

Whether it’s Micron scaling up its CAPEX, or Samsung and SK hynix expanding their HBM and associated production lines, establishing cleanrooms, acquiring equipment, optimizing packaging processes, and achieving stable yields all require significant time. While every company knows that money is lucrative right now, announcing expansion today does not mean they can deliver products tomorrow.

Secondly, the demand/requirements for 2027 have already been pre-booked/committed.

Samsung already mentioned multi-year binding contracts, but the conference call content reported by Reuters was even more direct: the potential supply-demand gap in 2027 might be larger than in 2026. This signal is very critical. It means that leading customers are not buying on a quarterly basis, but rather locking resources on an annual or even multi-year basis.

Third, this round is not just about HBM.

If it were only about HBM, one could still explain the situation as “advanced products are strong, but everything else is average.” However, Samsung, SK hynix, and Micron’s current consensus indicates broader tightness across DRAM and NAND, particularly for high-capacity server DRAM modules, AI-oriented eSSDs, and storage related to KV cache. As long as this diffusion continues, the cyclical trend will keep propagating outward.

Fourth, the stock market typically peaks ahead of earnings reports, but it does not peak before the overarching narrative (or storyline).

The current main narrative is “shortages, price hikes, locked orders, and insufficient expansion.” Only when the main narrative shifts to “expansion realized, peak prices, and customers no longer rushing to buy” will the stock price truly seem to have peaked.

Under what circumstances might it end early

There is also a possibility that it could reach its peak sooner.

If two or three of the following events occur simultaneously between Q4 2026 and H1 2027, I would significantly become more cautious:

Cloud vendors are starting to revise down their AI capital expenditure growth rate.
Leading memory manufacturers no longer emphasize shortages, but instead focus on CAPEX, depreciation, and yield ramp-up.
Standard DRAM / NAND prices flatten or even reverse before HBM.
Budgets for mobile phones, PCs, and enterprise IT cannot keep pace with high-priced memory, thereby cooling demand.
Macro-level factors such as recession, tariff escalation, energy shock, or local production disruptions in Korea.

The fifth point, in particular, should not be underestimated. The 2008-2009 cycle already proved that even if the semiconductor industry is strong, it cannot withstand a simultaneous decline in macroeconomic credit and demand.

What to Focus On: Not ‘How Much It Has Risen,’ But These Turning Points

If you care about when the cycle ends, I suggest watching these signals rather than looking at daily stock price colors.

Indicators to Monitor	Suggests if It Remains Strong	Suggests if It Starts Weakening
The description of supply/demand for 2027 in leading company earnings reports	The cycle is still extending	Leading companies start preparing for a slowdown
Capital expenditure and progress of new cleanrooms / packaging facilities	Supply is still catching up to demand	Could plant seeds of oversupply in the next 12-18 months
Pricing and delivery times for DRAM / NAND / eSSD	Tightness is spilling from HBM to broader sub-markets	Supply starts moderating/softening
Scope of Hyperscaler AI CAPEX	The end-market engine is still pressing the accelerator	The most important demand source of the cycle has eased up
Inventory and booking behavior	Customers are still scrambling for capacity	Customers shift from rushing to buy to waiting for goods

My own criterion for judgment is simple: As long as the industry continues to emphasize supply shortages rather than inventory cleanup/recovery, this cycle has not truly ended.

Conclusion

As of May 12, 2026, my conclusion is:

This semiconductor cycle is highly unlikely to end in 2026.
The massive surge in US storage and Korean semiconductors is due to one factor: AI has completely unlocked the profit elasticity of the memory complex.
The end mechanism of past cycles is not mysterious; the core elements are simply falling demand, rising inventory, and supply catching up.
The most probable critical period this time is neither now, but from the second half of 2027 to the first half of 2028.

Of course, this doesn’t mean we should rush into investing blindly now. How can I put it? For an industry like semiconductors, what they truly fear is never the economic “boom” cycle itself, but rather everyone believing that the boom will continue indefinitely, leading them to desperately overexpand capacity during high-profit years.

The cycle usually doesn’t die when it’s worst, but when it’s best, hottest, and most out of stock.

References

Writing Notes

Original Prompt

Recently, semiconductor-related stocks have been soaring, covering U.S. storage/memory and Korean semiconductors. I need to gather relevant information to predict when this current semiconductor cycle will end. Also, how did past semiconductor cycles conclude, and for how long did they last?

Summary of Writing Approaches

First, clarify why this rally is so strong before discussing when it will end; otherwise, the prediction lacks foundation.
The body article intentionally separates firsthand facts from my judgment, avoiding presenting time-sensitive information as permanent conclusions.
The historical section does not try to cover every small fluctuation but only focuses on the few major downturns most significant for investment.
The predictive section avoids making “fortune teller”-style single-point guesses on the peak; instead, it provides a baseline timeframe and trigger conditions for an early top.
This article intentionally did not elaborate on the independent cycles of wafer

Extended Brainstorming

Topic	Include in Main Text?	Reason
Analyzing NVIDIA and TSMC together	Partially Downplayed	They are important, but they will pull the theme from the memory cycle toward the entire AI industry chain, which is easy to lose focus on.
Expansion of domestic Chinese storage and mature processes	No	It affects supply, but this piece’s main thread is US memory/storage and South Korean semiconductors

Futu Securities cost price algorithm, which one is the default for Hong Kong brokers and domestic brokers?

Fri, 08 May 2026 23:45:20 +0800

Regarding this problem/issue from Futu, I will state the conclusion first.

Futu Securities currently defaults to displaying cost-averaging, which is not what many people understand as simply calculating an average based only on buys and ignoring sells. Extending this concept further, Hong Kong brokers do not have a unified default methodology. If it is a holdings page aimed at Chinese retail investors, common display methods include cost-averaging, average cost, and breakeven price (or principal protection price). However, if the broker is an international firm like IBKR that places more emphasis on tax lots and statement consistency, FIFO (First-In, First-Out) is generally the default method.

When discussing domestic brokerage firms, if we refer to the most common field in A-share trading clients—the “Cost Price/Holding Cost Price”—it is more commonly focused on the diluted cost basis or break-even price, rather than simply the average purchase price. This is just because different brokerages do not have standardized naming conventions; some call it “holding cost,” others call it “diluted cost,” and some provide a separate figure for the “average purchase price.”

Let’s first break down the three concepts

These three terms are often used interchangeably, but they are not the same.

Basis	Core Logic	Will it change after selling?	Common Usage
Average Cost	Looks only at purchases, ignores the effect of selling on remaining positions	Usually no	Viewing the pure average buying price
Averaging/Break-even Price Maintenance	Allocates historical gains/losses from sales back to the remaining position	Yes	Displaying “how far from break-even” on the holding page
FIFO (First In, First Out)	Sells what was bought first, matched by lot	Yes, but the logic is lot matching	Statements, taxes, realized/unrealized gains/losses

If you just want to see “what the approximate cost of my remaining stocks is,” average cost is the most intuitive.

If your concern revolves around calculating how far your current holding is from the break-even point after engaging in frequent, short-term trades on a stock, then cost averaging techniques are more directly applicable.

If you are concerned with knowing exactly which old position/lot was sold first, that is no longer just a front-end holding display issue; it becomes an underlying lot matching problem like FIFO.

What is Futu’s current default type?

I checked the Futu Hong Kong Help Center and the Futu NiuNiu Help Center on May 8, 2026, and obtained two definite facts.

Firstly, the Futu Help Center separately explained two algorithms: Amortization Cost Price and Average Cost Price.

\[ \text{Adjusted Cost Price}=\frac{\text{Total amount bought during holding period}-\text{Cash dividends}-\text{Total sales proceeds}}{\text{Quantity held}} \]

\[
\text{Average Cost Price}=\frac{\text{Previous Average Cost Price}\times\text{Quantity}+\text{Current Buy Price}\times\text{Quantity}}{\text{Total Holding Quantity after Purchase}}
\]

Futu NiuNiu’s “Holding Field Introduction” page directly labels the stock holding field as Cost Price (Diluted Cost). The formula provided on this page does not incorporate cash dividends, focusing instead on explaining the display logic of the holdings page. However, both the HK Help Center’s “Cost Price Introduction” and “FAQ” include cash dividends in their formulas. While the descriptions differ slightly, the core methodology is consistent: both reflect Diluted Cost, rather than simply the average purchase price.

Secondly, Futu writes more directly in its “FAQ”: “Futu currently uses the weighted average cost price.” It also explains a phenomenon that many users might be confused by: when the selling amount is greater than the buying amount, but there are still remaining positions in the account, the cost basis can become 0 or even negative.

This fact itself indicates that Futu’s default display is not based on “average cost,” but rather uses a metric that has incorporated/accounted for historical realized gains and losses from selling.

So, if the question is “What is Futu Securities’ default cost basis algorithm?”, the answer is quite straightforward: It defaults to average cost basis.

What standard type do Hong Kong brokerage firms default to, which is too complex to define simply?

Many people tend to treat “Hong Kong securities firms” as a single entity, which is an imprecise statement.

The few examples I found show that it is not uniform.

Brokerage / System	Method Found	Conclusion
Futu	Currently uses cost basis averaging	Default bias toward cost basis averaging
Tiger Brokers	Clearly distinguishes between average cost, FIFO, and cost basis averaging, and supports selection in settings	Not covered by a single default logic
IBKR	Tax lots default to FIFO, with the option to change the default match method	Default bias toward FIFO

Tiger Securities’ help center is quite informative. It separately lists Average Cost, FIFO, and Weighted Average Cost, and provides different results for the same set of trades under three algorithms. Furthermore, on the “Realized Profit/Loss” and “Unrealized Profit/Loss” pages, Tiger clarifies that the results differ between FIFO and average cost methods, and users can select them in the settings.

What does this mean?

The question of which type Hong Kong securities firms default to is highly likely to lead to biased conclusions if a specific brokerage firm is not first specified.

My judgment is as follows:

In the context of Chinese-language retail brokerage apps, such as Futu or Tiger, the common practice revolves around presentation metrics related to holdings, including averaged cost basis, average cost, and breakeven price.
For international brokers like IBKR, in reporting and tax contexts, FIFO (First-In, First-Out) is generally the more common default method.
Therefore, stating that “Hong Kong brokers default to FIFO” or that “Hong Kong brokers default to averaging cost basis” is inaccurate. The correct statement must specify both the particular broker and the specific page/context.

Which type is more commonly used by domestic brokers?

If we narrow the scope to the most common “Cost Price” field found on A-share brokerage client platforms, my conclusion is:

Domestic brokerages are more likely to focus on cost averaging / capital preservation, rather than simply buying at an average price.

Guoxin Securities wrote this definition very clearly a long time ago. Its “holding cost basis” algorithm is:

\[ \text{Average Holding Cost Price}=\frac{\text{Total funds used for purchases during holding period}-\text{Total funds obtained from sales during holding period}}{\text{Number of available shares}} \]

The key point of this formula is that the selling amount will inversely affect the remaining holding cost. This is no longer simply the average purchase price; it clearly leans closer to the idea of averaging down costs.

If we look at Industrial Securities and Yangtze River Securities, they are more granular in their definitions/metrics. They explain “average purchase price,” “cost basis of holdings,” “averaged cost,” and “break-even price” separately. Especially Yangtze River Securities, they directly state that their own cost types are divided into four categories.

This conversely shows two things.

Firstly, domestic securities brokers are not incapable of calculating average cost; rather, they often use the average cost as a separate field, and do not necessarily treat it as the default “cost price.”

Secondly, domestic brokerages pay more attention to displaying ‘how far the current holding is from breaking even’ in their front-end user interface, which makes it easy for them to incorporate the impact of selling into the displayed cost basis. This approach is actually closer to Futu, rather than IBKR’s lot accounting logic.

Why Many People Feel That “The Cost Price Is Wrong”

It is usually not that the broker calculated it incorrectly, but rather that your definition of “cost basis” in your mind is not the same thing as the “cost basis” shown on the brokerage platform.

The three most common errors are.

Type One: Interpreting based solely on the average buying price

I think that after selling off a portion, the cost basis of the remaining position shouldn’t change.

This is only valid based on the “average cost” metric/basis.

However, if the broker displays the adjusted/average cost basis, then the profits from your previous sales will indeed lower the cost of your remaining holdings; conversely, losses from previous sales will raise the cost of your remaining holdings.

Secondly, conflating the criteria for displaying holdings with the report criteria

The App’s holdings page might be showing you the averaged cost basis.

However, the monthly statements, tax forms, and allocation of realized gains/losses may be processed using FIFO or other lot matching rules.

These two metrics are not necessarily consistent.

Third kind: Corporate Actions, Portfolio Reallocation, Dividends, and Stock Splits

This kind of scenario is the most prone to miscalculating/confusing the cost price.

Futu and Tiger have warned in their Help Centers that, “Corporate Actions, and cost basis after transactions are for reference only and may be inaccurate.” This reminder is very important because many users treat the front-end display as the definitive accounting ledger, which can lead to greater confusion.

My Conclusion

To wrap up. / Let’s summarize the end.

If you are referring to Futu:

Futu Securities currently defaults to the average cost basis.

If you are referring to the overall picture of Hong Kong securities firms/brokerages:

No standardized default algorithm.
Futu tends to average/dilute the cost basis.
Tiger supports switching between multiple cost algorithms.
International brokers like IBKR’s default tax lot calculation leans more toward FIFO.

If you are referring to the entire domestic brokerage industry:

More commonly, it follows a focus on cost averaging / maintaining the initial principal cost.
The “average purchase price” usually exists, but it is often in a separate field and is not necessarily the default displayed “cost price.”

To really avoid misjudgment, the most reliable method is not to memorize broad conclusions like “which type of brokerage Hong Kong uses by default” or “which type of brokerage mainland China uses by default,” but rather to look directly at these three questions:

Does this page display the “Holding Cost,” “Average Purchase Price,” or “Break-even Price”?
Will selling transactions affect the cost basis of the remaining position?
Is this metric for front-end display, or is it a report/tax lot standard?

If you understand these three points, you generally won’t be misled by cost pricing discussions again.

References

Writing Notes

Original Prompt

Futu Securities’ cost basis algorithm; which one do Hong Kong brokers default to? Which one do domestic brokers default to?

Writing Outline Summary

First, separate Average Cost, Diluted Cost, and FIFO bases to avoid mixing concepts at the start.
Address the most specific question regarding Futu first, and then expand to Hong Kong brokers and domestic brokers.
For the section on Hong Kong brokers, do not provide a “one-size-fits-all” conclusion; instead, clarify the differences using three examples: Futu, Tiger, and IBKR.
In the section on domestic brokers, focus on explaining that “Default Cost Price” and “Average Purchase Price” are often not in the same field.
This article intentionally avoids detailing tax declaration procedures because the core issue raised by users concerns the standards for displaying holdings, not tax filing treatment.

Extended Brainstorming

Topic	Whether to include in main body	Reason
IBKR’s default FIFO	Included	Helps readers understand that “Hong Kong brokers” internally differentiate between retail holding displays and international tax lot calculations.
Tiger’s multiple algorithm switching	Included	Proves that the way “Hong Kong brokers” default cannot be generalized.
Whether all domestic brokers default to cost averaging	Not drawing a definitive conclusion	Evidence shows this is common, but different brokers have naming and implementation variations; it is unsuitable to state definitively.
Stock Connect cost basis	No	Charles River Securities explicitly warns that its cost price calculation is not applicable to Stock Connect; expanding on this will deviate from the main theme.
Tax filing and cost basis	No	Helpful for understanding the current issue, but it will shift the article from holding display issues to tax handling issues, becoming too tangential.

After AI stocks skyrocketed

Fri, 08 May 2026 19:40:15 +0800

The most unusual aspect of this current AI market cycle is not that Nvidia has risen sharply, but that the increase in value has been transmitted throughout the entire industrial chain: first GPUs, then servers, switches, ASICs, HBM, and finally to NAND, hard drives, power, and data centers.

If it were just a concept, the market trend shouldn’t last this long. But saying that it has already formed a complete profit cycle might be premature.

I prefer to view it as a “bull market driven by certain expenditures”: cloud vendors and model companies are genuinely spending money, and upstream companies are indeed collecting revenue, which is why stocks rose first; however, terminal applications have not yet proven that these investments can reliably generate enough profit, meaning the risk of a bubble also exists.

First, Clarify the Definitions/Scope

This is not investment advice. The stock price data uses an approximate retrospective based on public market quotes and historical closing prices, focusing on stages and logic rather than aiming for decimal precision for every trading day.

I set the starting point to 2022-11-30, which is near the date ChatGPT was released. The endpoint is based on an understanding of public market conditions around the writing time, 2026-05-08.

The gain can be roughly understood as:

\[ 
\text{Increase Rate}=\frac{\text{Period-end Price}-\text{Period-start Price}}{\text{Period-start Price}}
\]

There are two sources of potential error here: First, different websites do not handle adjusted pricing, stock splits, and intraday prices consistently; second, the US stock market on May 8, 2026, has not yet closed, so real-time prices will continue to change. Therefore, the main body focuses more on “multiple levels” and “relative strength,” rather than presenting it as a trading system.

Timeline: Where AI Goes, Stocks Rise

Phase	AI Development Status	Top Gainers	What the Market is Buying
Nov 2022 to Mar 2023	ChatGPT goes mainstream

May 2023 was the most critical turning point in the first phase.

When ChatGPT gained mainstream attention, the market could still question: Is this a chatbot bubble? However, Nvidia’s earnings guidance in May 2023 directly shattered that doubt. The data center revenue and next quarter’s revenue guidance were clearly higher than market expectations, marking the first time the market saw “model capability” translating into actual “GPU orders.”

This is why Nvidia did not begin its rise in 2024, but rather entered the major uptrend phase as early as 2023.

As of 2024, the market trend has shifted from merely “buying GPUs” to requiring the acquisition of entire “AI factories.” Training large models is not just about buying several graphics cards; the true expense lies in the complete cluster: GPU, HBM, networking equipment, servers, liquid cooling, power supply, data center space, and software stack. If any single component is missing, the system cannot function.

Therefore, Super Micro Computer will surge, Broadcom will surge, TSMC is expected to rise, and Oracle will also increase. They are not the same companies, but they all stand on the value chain of AI infrastructure.

After 2025, the market began searching for a second level of certainty: whether models could actually be integrated into enterprise workflows. Palantir’s AIP is representative of this phase; the market isn’t buying merely a software company, but rather the imaginative potential of “AI entering enterprise decision-making and operational systems.”

The speculative potential here is significantly higher. GPU companies are already generating revenue, while enterprise AI software is still proving its capacity for sustained revenue generation.

Approximate Gains of Major Companies

Judging by the trend since the release of ChatGPT, the stocks experiencing the most dramatic gains are not large-cap tech giants like Microsoft, Google, and Amazon, but rather companies with relatively small market capitalizations whose earnings potential has been amplified by AI.

What is most noteworthy in this table is that price appreciation is determined not only by the “degree of AI relevance,” but also jointly by the original market capitalization, profit elasticity, stock holding pattern, and the industry cycle.

Microsoft is certainly important, but it is simply too large. For Microsoft to rise 50%, the required capital and resulting increase in market capitalization are quite exaggerated. A small or medium-cap company, if suddenly perceived by the market as being on the main AI track, will find its stock price much easier to multiply several times.

This is also one of the core rationales behind the massive surge in flash memory.

Why Did Silicon Industry Surge So Much?

Solidigm is not a stock that has risen along with the AI trend since 2022. Its specialty is that, as of 2025, it spun off from Western Data to become a purer NAND and flash memory stock.

It skyrocketed, not just because “AI requires storage.” More accurately, it is due to several overlapping factors:

Factor	Impact
AI data centers require more high-performance storage	Training data, inference cache, vector search, data lakes, logs, and checkpoints all increase storage requirements
The NAND industry itself is undergoing a cyclical reversal	The storage industry has gone through a trough; after supply contraction, price recovery will be significant, leading to high profit elasticity
The listed target company becomes purer after going public independently	After being spun out from Western Digital, the market is easier to price based on the NAND/SSD cycle
Original market capitalization is not high	Compared to giants like Nvidia or Microsoft, less absolute capital is needed to boost the stock price
Short selling or low expectations are easily counterattacked	Once the earnings and guidance of a cyclical stock exceed expectations, valuation recovery can be very aggressive

I agree with the user’s assessment that the company has a low market capitalization and low capital expenditure, but I must add one point: Low market cap only provides potential elasticity, it is not the inherent catalyst for growth itself.

Without fundamental catalysts such as NAND price recovery, AI data center SSD demand, or improvement in financial performance following the company’s independence, a low market cap can only make it easier to speculate on, and also easier to fall back from. A truly significant market rally typically occurs when “low market cap + low expectations + marginal improvement in fundamentals” appear simultaneously.

Micron’s current movement appears to be driven by the powerful AI tailwind hitting the bottom of the storage cycle. The wind itself is enormous, and the market ground is perfectly dry.

Does this market cycle resemble the internet bubble?

Like, but not like.

The key point is that valuation precedes profit realization. The rise in many companies’ stock prices reflects expectations for the next 5 or even 10 years, as the market preemptively prices in companies with the potential to become infrastructure.

The difference is that: upstream companies in this round are already earning real money. Nvidia, TSMC, Broadcom, memory manufacturers, and server manufacturers are not selling PPTs; they are delivering hardware and services.

So, I do not quite agree with summarizing it in one sentence as “all bubbles.” It is more like:

Layer	Current Status	Risk
Compute Hardware	The profit loop is clearest, orders are real	If capex slows down, valuation and inventory will recoil/be hit by it
Cloud Infrastructure	Revenue is real, but depreciation and electricity cost pressure are high	Whether customers are willing to pay long-term for AI computing power
Enterprise Software	Has cases and growth, but ROI has not been widely proven	Many pilots, little scaling; easily shifts from enthusiasm to budget scrutiny
Consumer Applications	Many users, clear monetization divergence	The balance between customer acquisition, retention, inference costs, and subscription pricing may not be achieved

The biggest contradiction currently is that the AI upstream segment has formed a profitable closed loop, but the downstream applications have not.

Nvidia profits from cloud vendors, which spend capital expenditures (CapEx). This CapEx ultimately gets paid for by enterprise customers and consumers. If the end-users do not pay enough, or if AI fails to deliver sufficient cost reduction and efficiency gains to enterprises, someone in the chain will ultimately bear the depreciation and valuation pressure.

The stock price might not crash immediately, but the market will begin to ask a harder question: Where is the return on investment (ROI) for these GPUs?

How Do Research Reports and Institutions View the Risk of a Collapse?

The institutional views I found are not consistent, but they can be grouped into three categories.

The first camp is the cautious one. The title of the Goldman Sachs 2024 report on generative AI was very direct: it suggests that investment is too high and returns are too low. Its core argument is not that AI is useless, but rather questioning whether massive short-term capital expenditure can generate sufficient returns.

The second type is the moderate view. Sequoia has raised the issue of an “AI revenue gap”: to support GPU investments, the entire ecosystem needs to generate very large end-user revenues, but current application layer revenue has not yet caught up with infrastructure investment. This is not a bearish take on AI; it is a reminder that the commercial loop has not been closed yet.

The third category represents the optimists. They believe that AI will follow a path similar to cloud computing: first requiring years of infrastructure investment, followed by a gradual release of software and service revenues. This assessment also has merit; after all, cloud computing itself was questioned for its high costs in its early days.

The problem is that the stock market won’t wait 10 years to price it. It will buy early, and it will also kill early.

My judgment is:

The market may not face a sharp downturn in the short term because capital expenditure remains high, orders are still flowing, and the AI race is ongoing. As long as major companies like Microsoft, Google, Amazon, Meta, and Oracle continue to expand their data centers, the revenue of upstream hardware companies will remain supported.

But it will definitely undergo a rigorous ROI review in the mid-term. The triggers might be:

Cloud vendors slowing down AI Capex;
The large model price war results in insufficient inference revenue coverage for costs;
The failure rate of enterprise AI projects transitioning from pilot phase to production is too high;
High inventory levels in certain upstream links;
Interest rates or the macroeconomic environment make the market reluctant to assign high valuations to long-term narratives.

This does not mean AI technology has failed. After the dot-com bubble burst, the internet did not disappear. What truly disappeared was the portion that had been prematurely overvalued in the valuations.

Which metrics should I focus on

If I continue observing this round of market trends, I won’t just look at model announcement conferences/releases.

Several more useful indicators are:

Metric	Why It Is Important
CAPEX growth of four major cloud providers	Determines the sustainability of upstream hardware orders.
NVIDIA data center revenue and gross margin	Indicates whether compute power demand is genuinely strong or if prices are starting to ease.
HBM / NAND / SSD pricing	Shows whether storage market recovery is occurring, or if it

Especially the last one. AI cannot only look at revenue, but it must also account for depreciation.

Buying GPUs is not free, nor is building a data center. If AI service revenue growth is very impressive, but free cash flow becomes increasingly concerning, the market will eventually reprice it.

Conclusion

The surge in AI stocks this round started with the technological shock brought by ChatGPT, then moved to Nvidia’s orders, then to the entire AI factory, and finally spread to storage, memory, power, and enterprise software.

Cymbet’s surge is not an isolated event. It stands at the intersection of AI storage demand, NAND cycle reversal, independent listing, and low market cap elasticity; therefore, its gains will be more exaggerated than those of many mega-cap companies.

However, the more such market conditions arise, the more it is unwise to only look at “infinite AI demand.” Capital markets prefer to incorporate long-term trends into stock prices all at once, and they are also adept at correcting in reverse when the rate of realization is insufficient.

AI is probably not fake. The problem is that current stock prices have already assumed that AI will soon become a very profitable, very stable, and very large-scale business.

This default value is where problems are most likely to occur later.

References

OpenAI: Introducing ChatGPT, November 30, 2022.
OpenAI: GPT-4 research, March 14, 2023.
NVIDIA: FY2024 Q1 Earnings Report, May 24, 2023.
NVIDIA: FY2025 Q1 Earnings Report, May 22, 2024.
Western Digital: Completes SanDisk Spin-Off, February 24, 2025.
SanDisk: FY2026 Q3 Earnings Report, April 30, 2026.
Goldman Sachs: Gen AI: Too Much Spend, Too Little Benefit?, 2024.
Sequoia Capital: AI’s $600B Question, 2024.
Gartner: 30% of GenAI Projects Will Be Abandoned After Proof of Concept, July 29, 2024.
MIT NANDA: The GenAI Divide: State of AI in Business 2025.

Author’s Notes

Original Prompts

The AI boom has caused many companies' stock prices to soar. We need to organize the stock gains of related companies since ChatGPT was released according to a timeline, marking periods of steepest increases, what state of AI corresponded to those times, and why the corresponding stocks surged. Why did [Shandili/Company Name] surge so much? Besides the influence of AI, there are other factors. [Shandili]'s original market capitalization was not high; lifting it requires capital, and if the market cap is low, less capital needs to be expended. Search through research reports: Will this wave of AI eventually collapse? Currently, everything is burning cash, and there is no complete profit cycle established yet.

Writing Outline Summary

Instead of presenting all AI stocks as a market data list in the main body, it writes according to categories such as “model capability, order fulfillment, infrastructure diffusion, enterprise implementation, and storage cycle.”
The section on Flash Memory deliberately did not only focus on AI demand, but also included NAND cycles, independent listing status, and low market cap elasticity.
Regarding the bubble issue, it was not presented directly as collapsing or stable; instead, it was broken down into upstream profit cycles and downstream ROI cycles.
The article minimized details of many individual companies (e.g., AMD, TSMC, and power stocks), otherwise it would become a mere stack of data/information dumping.
Stock price metrics are limited to the multiples level, used for explanatory purposes, not for trading judgment.

Expanded Brainstorming

Why is Wuliangye causing such a commotion/stir?

Mon, 04 May 2026 14:18:43 +0800

Wuliangye’s performance fluctuation this time was not merely an ordinary dip; it directly disrupted the long-held unspoken norms within the baijiu industry. According to its 2025 annual report, the company reported full-year revenue of 40.529 billion yuan and net profit attributable to owners of 8.954 billion yuan. What is even more striking is that the company also conducted prior accounting error corrections for the first quarter, half year, and third quarter of 2025. Simply put, many figures from 2025 that initially looked impressive were later recalculated.

My judgment on this matter is straightforward: It’s not simply an “earnings crash,” but rather Wuliangye telling the market that their past approach of relying on channel pressure and reporting through future reserves can no longer be sustained.

What were the previous rules

The most core rule in the Baijiu industry used to be not “how much was actually sold at the terminal,” but rather “getting the product out first, and making the accounts look good first.”

Manufacturers push goods onto distributors, who then move them down to lower-level channels. This allows revenue and profit to be recorded on reports prematurely. As long as wholesale prices remain stable and inventory can still be absorbed, everyone assumes this logic is sound. On the surface, it appears that the brand is expanding; in reality, many times, it is the channel preempting future sales for the manufacturer.

This model is viable, relying on three prerequisites. First, the industry is still expanding, and distributors are willing to commit for the long term; second, the pricing structure remains stable, leading everyone to believe that the goods can eventually be successfully sold out; and third, brand strength is sufficient so that inventory buildup will not immediately turn into a crisis.

So previously, what the Baijiu industry feared most was not slow sales, but price instability (or “price erosion”). Once prices start to drop/are unstable, distributors lose confidence, and inventory shifts from being “good-looking on paper” to being “strikingly visible in the warehouse.”

Why Did Wuliangye Stand Out?

I think Wuliangye will stand out/step up this time, based on at least three reasons.

First, the industry has truly entered a period of deep adjustment. Wuliangye clearly states in its annual report that the baijiu sector will enter a comprehensive and deep adjustment phase—the “deep water zone”—in 2025, and the overall development landscape of the industry is undergoing accelerated evolution. Furthermore, it itself admits that it must proactively align with market adjustments and actively help alleviate pressure and difficulties within distribution channels.

Secondly, the former methodology is starting to undermine itself. It was stated very clearly in the preliminary accounting error correction announcement that the company reviewed its 2025 business model and, based on the principle of prudence (or conservatism), adjusted the accounting calculations related to recognizing some of the 2025 business revenue. In other words, those seemingly perfect reports previously generated by relying on channels and confirmation timing can no longer be taken as the truth.

Third, governance pressure is also increasing. The annual report summary stated that the Chairman was absent from the board meeting because he could not perform his duties normally. Furthermore, external media also disclosed at the end of February that custodial measures were taken against him. Given this situation, it is even less likely that the company can continue to postpone old issues; rather, historical burdens are more prone to being exposed all at once.

So, this time’s “shake-up” isn’t because Wuliangye suddenly became particularly assertive; rather, it’s that it can no longer avoid shaking things up. The old rules have been stretched to their breaking point, and continuing to maintain superficial prosperity will only make the underlying deficits much larger.

Changes Before and After Financial Reports

What is most noteworthy this time is not the total figure for any given year, but that the three sets of reports disclosed for 2025 have been revised simultaneously. The correction announcement clearly stated that the items adjusted pertain to parts of the consolidated balance sheet and consolidated income statement for Q1, semi-annual, and Q3 of 2025, and do not affect the cash flow statement.

First, look at the most core revenue and profit:

Period	Original Reported Operating Revenue	Restated Operating Revenue	Original Reported Net Profit Attributable to Parent	Restated Net Profit Attributable to Parent
Q1 2025	369.40 Billion Yuan	170.86 Billion Yuan	148.60 Billion Yuan	44.16 Billion Yuan
H1 2025	527.71 Billion Yuan	235.10 Billion Yuan	194.92 Billion Yuan	46.24 Billion Yuan
Q3 2025	609.45 Billion Yuan	306.38 Billion Yuan	215.11 Billion Yuan	64.75 Billion Yuan

This set of numbers is very striking. Revenue has been revised downwards three times, each time approaching a “halving.” The profit reduction is even more severe. Especially in the semi-annual report, net profit attributable to owners dropped directly from 194.92 billion yuan to 46.24 billion yuan—it almost completely stripped away the sense of prosperity that was previously there.

If you look at the changes in the balance sheet, it becomes clearer that these are not just minor definitional tweaks or cosmetic adjustments:

Period	Other Current Assets	Other Current Liabilities	Other Payables	Taxes Payable
Q1 2025	11.82 billion RMB -> 34.99 billion RMB	5.04 billion RMB -> 190.93 billion RMB	115.30 billion RMB -> 106.16 billion RMB	81.68 billion RMB -> 56.83 billion RMB
H1 2025	1.91 billion RMB -> 81.87 billion RMB	4.23 billion RMB -> 277.49 billion RMB	189.05 billion RMB -> 170.09 billion RMB	45.40 billion RMB -> 39.31 billion RMB
Q3 2025	1.44 billion RMB -> 85.47 billion RMB	3.86 billion RMB -> 278.27 billion RMB	93.46 billion RMB -> 74.49 billion RMB	15.56 billion RMB -> 15.44 billion RMB

I am more concerned about “Other Current Liabilities.” Its elevated value at all three disclosure points suggests that some items from the past were indeed masked within the channel and settlement cycle/process. In other words, this isn’t simply a decline in profit; rather, both revenue recognition and the underlying channel strategy are being revealed simultaneously.

When viewing the full-year 2025 report alongside

What Comes Next

In the short term, the baijiu industry will look worse, but also more realistic.

Based on the restated Q4 2025 and Q1 2026 reports, the statements will be more “straightforward” than before, but also tougher to look at.

More importantly, the industry will gradually shift from focusing on “who has better distribution power” to determining “who can genuinely sell the wine directly to the consumer.” Wuliangye has outlined this direction in its annual report: traditional channels, e-commerce, group buying, direct sales to corporate clients, end-point optimization, and consumer nurturing—all point towards achieving more direct sales movement.

The road ahead won’t be easy. Once the old rules are dismantled, distributor inventory pressure will become more pronounced, manufacturer profit fluctuations will increase, and the intermediate players in the industry that rely on price differences and information asymmetries for revenue will also face greater difficulties.

What I Pay More Attention To

What is truly worth watching regarding this matter is not how much Wuliangye dropped this time, but whether it has the guts to admit: that many of the impressive figures over the past few years were never built upon genuine consumer spending.

If we view white liquor as a mature industry, it is highly unlikely that it will revert to a state characterized by uncontrolled expansion, artificial supply suppression, or unjustified price hikes. The sector will undergo deeper differentiation; while leading brands will maintain their position, growth will slow down, and corporate financial reports will reflect genuine business performance rather than speculative narratives.

For investors, this is not pure bad news. The drawback is that the high-growth era, which was sustained by the old order, is unlikely to return; but the upside is that when looking at financial reports in the future, we won’t constantly have to suspect whether goods were stuffed into channel warehouses again.

References

Writing Notes

Original Prompt

Analyzing the causes and consequences of Wuliangye’s poor performance. What were the previous industry norms? What is the subsequent development? And why did Wuliangye challenge the status quo?

Summary of Writing Concepts

First, lock down the 2025 report and the prior accounting adjustments to demonstrate that this is not normal volatility, but rather a recalculation of metrics definitions.
Restrict the concept of “old rules” to the existing logic: channel inventory pressure, immediate booking upon shipment, and using inventory for reporting purposes.
Explain Wuliangye’s actions as the result of being pushed by industry adjustments, channel pressures, and corporate governance pressures simultaneously.
The focus later should be on how the industry will move toward actual consumption/real sales, consumer centrality, and more direct distribution models.
This article deliberately avoids a cross-comparison between Wuliangye and Moutai, nor does it include short-term stock price predictions; the main thread focuses solely on changes in rules/regulations.

When Alipay buys 006327, which day's net value is it calculated on?

Mon, 27 Apr 2026 23:28:12 +0800

I bought 006327 today on Alipay. I thought I would place the order before 3 PM, but what I received was based on “today’s Hang Seng Tech closing price.” The profit displayed on the page didn’t update for ages, and confirming the shares was also delayed. My first reaction was sheer shock: how is this thing actually settled/transacted? And why do I have to wait another two days?

To be honest, this misunderstanding is extremely common. Alipay has made buying and selling over-the-counter (OTC) funds look too much like placing stock orders, but fundamentally, there are two pitfalls in this matter. First, 006327 is absolutely not the Hang Seng Tech Index Fund. Second, when you buy a fund through OTC channels, the price you get is not based on an index’s real-time closing point, but rather the Net Asset Value (NAV) calculated by the fund company for that specific day. Furthermore, coupled with the

Correcting a Misconception

006327 is Yifangda CSI Overseas Internet 50 ETF Feeder (QDII) A. The underlying index it tracks is the CSI Overseas China Internet 50 Index, not the Hang Seng Tech Index.

These two things appear to be associated with the Hong Kong internet sector, and the market often rises and falls together, making it very easy for them to be mixed up into one concept. However, when it comes down to actually placing an order, the code won’t care about your feelings; if you buy incorrectly, it is just wrong.

That’s also why you shouldn’t only look at the trend chart on Alipay when buying funds. What you are thinking of is Hang Seng Tech, but what you actually order/trade is 006327. Even if the profit display later shows no problem, your underlying reference point has already deviated.

The Index Closing Price Is Not What Truly Drives Trades

Off-platform fund subscriptions operate under a simple rule: the Unknown Price Principle. When you submit an application, you do not know what the final transaction net value will be; the actual price used is the Fund Share Net Asset Value (NAV) of this fund, which is calculated after the market closes on the day of your application.

The point of greatest confusion here is that many people treat funds like stocks. With stocks, you buy by watching the order book, and the price is instantly matched; funds are different. What you are buying is a share corresponding to a basket of assets. The value of this share must wait until the fund company settles the positions, exchange rates, cash holdings, and other elements according to rules before calculating that day’s Net Asset Value (NAV).

So, buying before three PM, provided that Alipay records this application as valid for the current day, then what you are locking is the application date, not “a specific index’s closing point at a certain time today.” It is also not simply taking the Hang Seng Tech closing price multiplied by a coefficient to execute the transaction for you.

Why Does QDII Always Make People Wait Two Extra Days

The confirmation process for open-ended funds is slower than stocks, and ‘QDII’ is even slower. The reason isn’t mysterious; it’s simply that they invest in overseas markets, which leads to a longer valuation chain.

According to the fund contract for 006327, the net value for day T is calculated on T+1, with announcements and confirmations completed by T+2. From a user experience perspective, this means that if you execute a purchase today, you will not see results today; tomorrow, you are highly likely still waiting, and it only feels “truly finalized” the day after tomorrow.

Many people might misunderstand this point, assuming that “the return is displayed two days later” means that returns for those two days were not calculated. This is incorrect. The more accurate statement is: The display of fund shares and net asset value results are delayed; it does not mean the fund only started calculating/running for you on the third day.

Delays in display are not the same thing as delays in effective execution/settlement. Specifically, for products like QDII, the final set of numbers displayed by the platform will inherently be approximately two working days later than when you press the buy button.

Examine the transaction(s) from April 27, 2026

If you submit a buy order for 006327 on Alipay before 15:00 on Monday, April 27, 2026, and this order is normally recorded by the system as a valid application for that day, then the approximate timeline would be:

2026-04-27: Submit application; the trade date is locked on this day.
2026-04-28: The fund company calculates the net asset value corresponding to this day (2026-04-27).
2026-04-29: According to the contract specifications, share confirmation and net asset value announcements usually occur around this date.
2026-04-30: On distribution platforms like Alipay, holdings and return displays are generally more complete and stable.

If there are any delays encountered due to overseas market closures, delayed currency valuations, or lag in the platform’s own display synchronization, this timeline may be subject to further postponement.

Therefore, your original understanding needs to be revised into a more accurate version:

Correct Part: The return display being delayed by two days is generally normal.
Incorrect Part: You did not buy based on today’s closing price of the Hang Seng Tech Index.

Conclusion

This seems like a page display issue, but fundamentally, you have not clearly separated the trading object from the trading mechanism. What you are buying is fund shares, not index points; and what you are placing is an off-exchange subscription application, not an on-exchange real-time transaction order.

When you encounter QDII products on Alipay again, what needs confirmation first is not “where today’s candlestick closes,” but rather three things: whether the code is correct, whether the underlying asset being tracked matches your intended purchase, and whether the platform recorded the application within the current trading day.

Don’t treat funds like stocks, and don’t assume all Hong Kong internet stocks are equivalent to the Hang Seng Tech Index. The former can cause you to misjudge the transaction price, while the latter might give you a biased view of what you’re actually buying. After going through all this fuss, what you should focus on isn’t the profit and loss numbers two days from now, but rather what exactly you purchased at the moment of placing the order.

References

E Fund China Securities Index Overseas Chinese Internet 50 ETF Connection Fund Contract (Official PDF)
Every Day Fund: 006327 Product Page
[Every Day Fund: 006327 Basic Information](https://fundf1

Author’s Notes

Original Prompts

Did I buy fund 006327 on Alipay today, purchasing it before three o’clock? Did this use today’s closing price of the Hang Seng Tech Index? Also, is my understanding correct that the profit display requires a two-day delay?

Continue, and then finalize into a draft.

Writing Approach Summary

First, correct the misidentification of the fund’s underlying asset, avoiding treating `006

Moutai's Net Profit Drops for the First Time, and it's Not Just Because Young People Aren't Drinking Baijiu

Wed, 22 Apr 2026 01:10:41 +0800

This matter deserves separate discussion because it directly impacts how we view Maotai. Previously, many people treated Maotai as an eternally rising consumption myth, and whether young people drank it or not was just a minor factor. Now, that is no longer the case. It is true that young people naturally have little interest in Baijiu (Chinese liquor), but this is more like a slow-moving variable. The sudden turnaround reported in the annual report appears to be driven by the contraction of an entire old system: obsolete business demands, traditional wealth distribution methods, and established status-driven consumption patterns.

It’s Just Superficially That Young People Don’t Drink

I used to casually attribute the cause to generational shifts, thinking that because young people were drinking craft beer, cocktails, and whisky, baijiu would naturally become marginalized. This conclusion isn’t necessarily wrong, but it’s insufficient to explain the sharp changes seen in Maotai’s latest financial report.

The reason is simple: generational preference changes are gradual; they won’t suddenly crash profits like this in a single quarter. What truly determines the marginal pricing and profit of high-end baijiu is never just whether “people drink it,” but rather “who buys it, what scenario they are in, and why they are willing to pay two thousand or three thousand yuan per bottle.” Often, high-end baijiu sells not its taste itself, but social efficiency, status affirmation, and relationship lubrication. When you are actually at a gathering, many people don’t care if this particular cup is worth the money; what they care about is whether the entire occasion matches this price point.

Kweichow Moutai itself has quite clearly articulated this change. At the national dealer gathering held on 2025-12-28, company management directly mentioned that “the demand from core traditional consumers is weakening,” while also emphasizing the need to find new customer groups and usage scenarios. This statement contains a great deal of information. It is tantamount to admitting that the core buyers who previously underpinned Moutai’s pricing structure are no longer as strong as they once were.

Why is Moutai like a shadow stock of real estate?

Here, by calling it a “shadow stock,” I don’t mean that Moutai’s reports contain a pile of real estate projects, nor do I mean that if real estate drops, Moutai will drop in perfect lockstep. I am more inclined to understand it as a reflection on the demand structure.

Over the past twenty years, many of China’s most prominent, sustained, and extravagant business scenarios were related to the real estate chain. Local government finances rely on land sales; municipal investment vehicles revolve around land assets, developers maintain high turnover rates, and general contractors, subcontractors, building materials, home decoration services, agents, and financial support systems all benefited immensely together. Once this chain was prosperous, business banquets, festival gifting, relationship maintenance, and project coordination would all be significantly amplified. The money wasn’t gradually chipped away from residents’ daily consumption; rather, it flowed out of a system characterized by high leverage, high turnover, and high premium pricing.

In this environment, Maotai is not just liquor; it functions more like a high-end social status symbol. It possesses several characteristics that make it uniquely suited for this role: a powerful brand, an elevated price point, strong recognition, and excellent circulation. Often, whether a bottle costs two thousand or three thousand (yuan), does not alter the budget logic of the gathering; rather, the higher the cost, the better it executes the signal transmission of “I value you,” “I am capable,” and “this event has prestige.”

So, I largely agree with the statement that “Maotai is a shadow stock of real estate,” but to elaborate fully: it does not allude to the houses themselves, but rather the high-end business consumption order created during the era of real estate credit expansion.

This existing order is in decline. The national real estate data disclosed by the National Bureau of Statistics for 2025 is also worrying: Real estate development investment fell year-on-year (YoY) by 17.2%, newly built commercial housing sales area declined by 8.7%, and capital secured by property developers dropped by 13.4%. When the real estate chain contracts, what often disappears first is not basic necessities like food and drink, but rather consumption that has the greatest budget flexibility, prioritizes prestige, and relies heavily on projects and connections to operate. High-end baijiu (Chinese liquor) is exactly caught in this predicament.

What This Annual Report Truly Reveals Is Not Just Weak Performance

The more crucial signal, actually, is that Moutai itself has started changing its narrative.

2026-01-09, Moutai publicly stated at a distributor meeting that “market transformation has become an unavoidable question for Moutai.” It also pointed out thoroughly that the past sales sector’s “non-commercialization” made it difficult for ordinary consumers to purchase wine fairly, conveniently, and genuinely. This statement is quite strong. It suggests that the company is aware that in the previous system, a large portion of the value did not come from genuine consumption needs, but rather from scarcity, artificial pricing structures, status group allocation, and channel premiums.

In other words, what Maotai is currently doing fundamentally means it is attempting to shed its “financial attributes,” reduce its dependency on gift-giving consumption, and dismantle its historically closed consumer segments. By moving toward ‘iMaotai’, direct-to-consumer channels (D2C), and emphasizing service and genuine needs, this effort is not merely channel reform; it is about establishing a new pricing foundation for the next stage.

This also explains why the company could still maintain positive growth during the first three quarters of 2025, but then suddenly turned negative for the full year. Previously, it was buoyed by momentum, brand recognition, and channels, but it could not sustain that through the fourth quarter. Old demand is declining, and new demand has not fully taken over/picked up, so naturally, the profit statement looks poor in the interim.

Rethinking This Stock

I don’t think Moutai is done for just because of this. Its brand, region, process, and supply constraints are still there; the moat doesn’t crumble overnight. There’s another important data point in the annual report: the company expects cumulative cash dividends of 650.33 billion yuan in 2025, which accounts for 79% of net profit attributable to owners. This is already very much like a cash cow machine.

But the challenge also lies here. Previously, the market was willing to give Moutai a high valuation because the core premise was that it was both a branded consumer product and operated like a highly predictable growth engine. Now that the growth component begins to weaken, the valuation anchor shifts in another direction: moving away from a high-growth consumption leader, and slowly drifting toward defensive assets characterized by lower growth rates, high dividends, and strong cash flow.

This is not the same thing for investors.

If you still look at it with the mindset of “earnings will always be double digits, wholesale prices will always be stable, and business demand will always be there,” it can be difficult to reconcile. Because the most lucrative demand from the old era truly declined along with real estate. But if you view Moutai as a super cash-flow asset that is going through de-bubbling, anti-speculation measures, and returning to its true consumption base, then it operates under a completely different valuation logic.

In Conclusion

Therefore, my current understanding of this matter is quite different from what it used to be.

It is true that young people are not drinking baijiu, and it is also true that the culture of celebratory drinking has strong historical cyclical characteristics. However, the more critical point is that the era which placed high-end baijiu on a pedestal was never sustained by ordinary people drinking it in daily life. Instead, it was propped up by an entire system: real estate credit expansion, ballooning corporate banquets (business entertaining), and distorted consumption driven by status/face. After the decline of the property market, baijiu will obviously not disappear immediately. But the segment of demand that is most profitable, least concerned with cost-effectiveness, and indifferent to high prices is no longer as robust as it once was.

For Moutai’s stock, the 2025 annual report will likely not be merely an ordinary performance fluctuation, but rather the beginning of a shift in its valuation narrative. Going forward, we cannot rely solely on brand loyalty; we must also assess the speed of genuine consumer uptake and determine whether channel reforms can successfully transform the past demand—which was propped up by temporary mechanisms/schemes—into a more sustainable consumption foundation.

References

Author’s Notes

Original Prompts

Analyzing the A-share listed baijiu company Maotai’s 2025 financial report data reveals a decline in net profit for the first time. Previously, the understanding was that young people don’t drink baijiu, and China’s so-called “drinking culture” is mostly a historical product. The latest understanding suggests that this historical product is real estate, and Maotai baijiu belongs to the shadow stocks of real estate. Since real estate has now declined, corresponding business banquets are no longer frantic, meaning this derivative product (baijiu) is not as needed anymore. Baijiu’s net profit remains high; for business banquets, they don’t care if the price is two thousand or three thousand—what they want is the “face” (prestige).

Writing Outline Summary

Use the summary of Moutai’s 2025 annual report, disclosed on 2026-04-17, to firmly establish “the initial decline in profit” as the key trigger point.
Instead of stating the main judgment as “young people don’t drink baijiu,” focus on the structure of business demand behind high-end liquor.
In the middle section, use three sets of official real estate data (development investment, sales area, and available funds) to explain why the real estate downturn will first hit high-end banquet budgets.
Separately added

XiaoMi’s “New and Old Replacement” and defensive battle with the electric vehicle sector

Fri, 06 Feb 2026 00:21:07 +0800

Worrying Mindset: Holding onto stocks persistently, observing calmly like a Buddhist, and paying attention to the fulfillment of “ecosystem premium.”

I. Market Overview: From 2025’s “Mania” to 2026’s “Consolidation”

2025 was a strong year for the Hong Kong stock market, with the Hang Seng Tech Index rising by 23.45% throughout the year – its best performance since inception. However, as we entered January 2026, the market transitioned into a clear pattern of “two upward trends and one pullback.”

Xiaomi stabilized above the HK$40 mark in November last year primarily due to better-than-expected Q3 earnings and strong buybacks by Lei Jun and the company. However, recent share prices have experienced a certain degree of retraction, currently fluctuating between 35-38 HKD, with the market digesting two key factors: the redesign period for the SU7 and sales overruns for the Ultra high-end vehicle model.

II. Xiaomi’s Recent Volatility: SU7 Ultra’s “High Price, Few Buyers” and YU7’s “Cornerstone”

Product Line Disconnect:

Recent data shows that Xiaomi SU7 Ultra’s sales plummeted to double digits (45 units) in December 2025. This demonstrates that ultra-luxury models (over $500k) are more a symbol of the brand than a sales foundation.
SU7 Refinement Pressure: Currently, SU7 is in a refinement transition period, with the next generation expected to debut in April. This “new vs. old replacement” has resulted in approximately a 22% month-over-month decline in January deliveries and consequently put pressure on the stock price.

YU7’s Ecosystem Premium:

Fortunately, the SUV model YU7 (Xiaomi’s second car) performed exceptionally stably, consistently topping the mid-to-large SUV sales charts for several months. This confirms my previous notes regarding “ecosystem premium” – MIUI members’ conversion rate for family vehicles (SUVs) is significantly higher than that for pure performance sedans.

The “Floor Price” Support from Repurchase:

Xiaomi recently repurchased 4.2 million shares for a total investment of RMB 1.5 billion. This “rising stock price, immediate buyback” strategy has established a psychological “leftmost safety margin” for the stock price.

III. Horizontal Comparison: The “Collective Cooling” of the New Energy Vehicle Sector

Looking at the comparison, Xiaomi’s 39,000 deliveries in January performed reasonably well during a period of industry decline (January is typically a seasonal low):

BYD: January new energy vehicle sales reached 210,000 units, with year-on-year and month-over-month declines of approximately 30%.
XIAO & LI (Weilai): Similarly facing delivery fluctuations, particularly in the pure electric sedan market, due to excessive competition, companies were resorting to price cuts to maintain market share.
Comparison Conclusion: Xiaomi’s volatility was primarily an “internal” product rhythm adjustment, rather than a fundamental collapse. Compared to other automakers relying solely on price wars, Xiaomi leveraged its “human-vehicle-home ecosystem” moat thanks to its smartphone+car+appliances business model, resulting in a shallower stock decline compared to pure automotive stocks.

IV. Vertical Comparison: HSTECH Index (HSTECH)

Index Performance: The HSTECH index rose 3.67% in January, with the overall price center rising.
Xiaomi vs. Index: Xiaomi has lagged slightly behind the broader market over the past two weeks. This is due to investment banks like Morgan Stanley lowering their target prices (to HKD 38) based on concerns that it will take time for mobile phone gross margins to recover.
Strategic Reflection: As I wrote in November last year, liquidity in Hong Kong stocks remains crucial. During a broad market rally, Xiaomi, as a blue chip stock, is constrained by profit-taking; but during a market correction, its buybacks and cash flow serve as the best defensive position.

V. Summary and Future Operations: Wait for the “April Showers”

Holding Logic: Given that the 2026 sales target is set at 550,000 units (as just announced by Lei), the current decline is more like a “dull knife cutting flesh” bearish correction, purging short-term speculators.
Focus Points: The pricing and configuration of the revamped SU7 in April, as well as whether the YU7 can continue to take over.
Action: Still no action, lying flat. Selling out at this position (35-38 range) is unnecessary; adding to positions hasn’t yet reached my “absolute safe haven” level.

Hang Seng Index Crash, Meituan Considering Exit? Examining Xiaomi, Alibaba, and Meituan’s New Valuation Anchors Through “Ecological Premium”

Tue, 18 Nov 2025 21:50:44 +0800

Recently, Xiaomi’s third-quarter financial report was released, with little expectation beyond the usual. The US situation is unsettled, Hong Kong stocks have recently experienced liquidity stagnation, and the Hang Seng Technology Index has retreated significantly.

As a major participant in the food delivery war, Alibaba, which bears the highest e-commerce taxes, hasn’t generated much discussion online.

I don’t remember if I wrote any previous articles, when I bought Xiaomi, I didn’t really think about what supported Xiaomi’s market capitalization at that time – perhaps after watching too many TikTok-related videos?

Recently, the Hang Seng Technology Index has been falling, and electric vehicle stocks in Hong Kong are also declining. Xiaomi has multiple buffs stacked on top of each other, with negative news flying everywhere, and its stock price has naturally fallen significantly.

Recently, it’s also been changing the head of its public relations department – conspiracy theories: they’ve been trying to change it for a long time, but the stakes are too high, and there are still many KOLs involved; if they suddenly make some wild claims, it will be difficult to resolve the situation.

This article primarily analyzes the valuation logic of Xiaomi Group and the market and policy challenges faced by Meituan and Alibaba recently.

The original draft contained too much complex content, collecting information from various sources and edited based on AI.

Xiaomi: The Logic Behind the Surge in Valuation (Platform Premium)

Core Argument: The capital market has given Xiaomi a surge in valuation, driven by its “platform-type company” and “ecosystem empowerment” premium – something that vertical manufacturers (like Xiaopeng, Zotye) lack.
Valuation Anchor Differences:
- Xiaomi: Anchored on the scarcity of “technology platforms” and “human-vehicle-home ecosystem.” The automotive business is viewed as a crucial step in ecological expansion, with risks dispersed across mobile and internet businesses.
- Automotive Manufacturers (Xiaopeng/Zotye): Anchored on “vehicle sales” and “unit profitability,” with valuation curves directly linked to the automotive industry cycle.
Ecosystem Empowerment Value: Xiaomi possesses a massive, highly engaged user base of over 100 million, with extremely high order conversion efficiency from cars, and low marketing costs. Its seamless “human-vehicle-home” synergy is viewed by the market as superior to the industry average differentiation.
Market Expectations: The market is buying into Xiaomi’s potential to successfully enter the trillion-dollar new “track entry ticket” and “ecosystem synergy value” in the next decade, reshaping its overall valuation logic.

Meituan: Considering Exit

Initially, the phrase “capable of bringing together all things, deeply ingrained in people’s hearts” was seen as a strong moat. However, it now appears that this moat isn’t as deep as initially thought.

Current Situation: When buying, little consideration was given; recently, the stock price has fallen, and a decision on whether to “cut losses” will be made after reviewing the third quarter financial results.

Alibaba (E-commerce Tax Policy Impact)

Tax Policy Changes: China’s recent “e-commerce tax” is primarily focused on two areas:
1. Cross-border E-commerce Retail Import Duty (“Haotai”): The preferential exemption for items below 50 RMB has been cancelled, leading to an approximate 11.9% price increase for haotai goods and driving industry compliance.
2. Domestic E-commerce Tax Information Reporting Standards: Platforms are now required to report sales information from merchants, effectively preventing C2C individual sellers who previously hid income from continuing, marking the entry of the industry into a core reshaping phase focused on compliance.

Impact on Platforms:

Taobao (C2C): May eliminate small businesses unwilling to comply in the short term, but in the long run, it will improve platform quality and consumer trust.
Tianmao/Jingdong (B2C): Long-term benefit, as these large B2C platforms and merchants are already regular tax payers, and this new policy creates a fairer competitive environment for them.
Reasons for Merchant Losses: Losses are concentrated in merchants originally relying on tax evasion to gain price advantages. For merchants eligible for small microenterprise preferential policies, the tax burden should be low or exempt after compliant reporting. Being eliminated is an inevitable step towards fair competition.

Big Tech Dominance in the U.S. Stock Market Intensifies: The Top 10 Companies Account for 40% of Market Capitalization, Is AI a Bubble or a Revolution?

Mon, 03 Nov 2025 23:01:27 +0800

Analysis of NVIDIA’s “Play” Worth a Billion Investment The global capital market is witnessing an unprecedented wave of centralization in 2025, centered around artificial intelligence (AI). This narrative not only reshapes the tech industry’s landscape but also exacerbates wealth inequality on Wall Street. The former “Magnificent Seven” no longer adequately describes today’s dynamics; the market is now dominated by a handful of super winners.

This article will delve into three key questions:

What proportion of the entire stock market do the top ten U.S. companies represent in terms of market capitalization?
Does AI constitute a bubble? Is NVIDIA’s (NVIDIA) multi-billion dollar “reciprocal investment” with OpenAI justified?
What are NVIDIA’s recent investment moves, and what is the underlying strategic logic behind them?

Write article: What proportion of the entire stock market do the top ten U.S. companies represent? Does AI constitute a bubble? Are NVIDIA’s investments and reciprocal investment with OpenAI justified? Compile NVIDIA’s recent investment actions and analyze their rationale.

Market Concentration: Top 10 Giants Dominate 40% of US Stock Market Capitalization

Based on the latest data from October to November 2025, the concentration in the US stock market (represented by the S&P 500 Index) has reached a remarkable level.

The top 10 US companies in the S&P 500 currently account for approximately 40% of the index’s total market capitalization.

This figure represents an even greater increase than the roughly 30% held by the “Big Seven” a year or two ago. As of early November 2025, the top 10 US companies were approximately:

NVIDIA: Market Cap Approaching $5 Trillion
Apple: Market Cap Approximately $4 Trillion
Microsoft: Market Cap Approximately $3.85 Trillion
Alphabet (Google): Market Cap Approximately $3.4 Trillion
Amazon: Market Cap Approximately $2.6 Trillion
Broadcom: Market Cap Approximately $1.75 Trillion
Meta Platforms: Market Cap Approximately $1.63 Trillion
Tesla: Market Cap Approximately $1.52 Trillion
Berkshire Hathaway: Market Cap Approximately $1.03 Trillion
JPMorgan Chase: Market Cap Approximately $0.85 Trillion

This high concentration means that the stock price performance of a few key companies can determine the entire market’s direction. And the core fuel driving this “super-concentration” is, without question – Artificial Intelligence.

AI Bubble and the $100 Billion “Closed Loop”: NVIDIA’s “Mutual Investment” with OpenAI

The “mutual investment” between NVIDIA and OpenAI, which you mentioned, became a global shock event by September 2025. This wasn’t traditional VC investment, but a strategic partnership valued at $100 billion, representing a significant shift in the AI landscape.

1. The “Closed Loop” Structure of the Partnership:

NVIDIA Investment: NVIDIA announced it would invest up to $100 billion in OpenAI.
OpenAI Procurement: OpenAI will utilize this funding (and other financing) to purchase NVIDIA’s chip systems, committing to deploy at least 10 ExaWatts (EW) of computing power – involving millions of GPUs – and adopting NVIDIA’s next-generation “Vera Rubin” platform starting in 2026.

2. Is This Model Reasonable?

This is a highly controversial topic, with market opinions divided between “revolution” and “bubble.”

Perspective 1: Reasonable - It’s the Inevitable Choice of the ‘AI Revolution’ (The Revolution Case)

For OpenAI (the Buyer): The path to Artificial General Intelligence (AGI) requires near-unlimited computing power. In today’s market, NVIDIA’s GPUs are the most powerful and have the most mature ecosystems – they are the “shovels.” Locking down supply of top-tier chips for several years is the only way to ensure its continued leadership in the AI arms race.
For NVIDIA (the Seller): OpenAI is its largest and most important customer. By investing $100 billion and deeply binding itself with OpenAI, NVIDIA effectively secures a massive order for its next-generation platform (Vera Rubin), guaranteeing future revenue and absolute market dominance over several years.

Perspective 2: Unreasonable - It’s a Classic ‘AI Bubble’ Feature (The Bubble Case)

“Closed Loop Financing”: Critics argue this is a “left hand, right hand” capital game. NVIDIA gives money to OpenAI, which then uses that money to order from NVIDIA. This creates massive revenue and growth on paper, but significantly amplifies valuations and bubbles.
Significant Financial Risk: Reports indicate NVIDIA was even discussing providing OpenAI with loan guarantees to build data centers. This means if OpenAI’s business model fails in the future (e.g., AGI doesn’t materialize or operating costs are too high leading to bankruptcy), NVIDIA faces risks of tens of billions – or even hundreds of billions – in debt.
Raises Antitrust Concerns: This transaction is viewed by regulators and competitors as “anti-competitive.” Analysts worry that, as a market “arms dealer,” NVIDIA will use its investment to deeply bind itself with the largest “mercenary” OpenAI, incentivizing it to refuse selling chips to OpenAI’s competitors (like Anthropic, Google, etc.), or offer worse terms, thereby stifling innovation.

Conclusion: This model is NVIDIA’s high-risk, high-reward strategic “play” to ensure its AI dominance. It’s both a catalyst for massive AI development and potentially the footnote on the largest bubble in history.

Nvidia’s Investment “Empire”: A Calculated Demand Creation

Nvidia has long been more than just a “chip vendor.” Through its venture capital arm (NVIDIA GPU Ventures), it is actively transforming into the “banker” and “ecosystem builder” for the entire AI gold rush.

📈 NVIDIA’s Recent Key Investment Actions

NVIDIA’s investment pace has accelerated dramatically in 2024-2025. As of October 2025, NVIDIA has invested in 59 AI startups this year, surpassing the 55 invested in throughout 2024.

Its investment portfolio covers almost every vertical within AI, with a focus on:

OpenAI (Foundation Models): In September 2025, it reached a strategic investment and procurement agreement valued at $100 billion.
Poolside (AI Programming): Revealed to lead an investment of up to $100 million, supporting the development of AI software engineers in October 2025.
Nokia (6G/Telecoms): Announced a strategic partnership and invested $100 million jointly developing AI-RAN (Radio Access Network) to capture the 6G market.
Figure AI (Humanoid Robots): Investing in this star robotics company, positioning itself for the future of “Embodied Intelligence.”
Perplexity AI (AI Search): Investing in this OpenAI competitor, securing a foothold in the next-generation information gateway.
Wayve (Autonomous Driving): Investing in the UK’s autonomous driving company to ensure its dominance in automotive chips.

🤔 Investment Logic Analysis: Why is it Reasonable?

Nvidia CEO Huang Renjun’s strategy is very clear. The core logic of its investments isn’t purely financial returns, but rather to serve its core chip business, which can be summarized into three key objectives:

1. Core Objective: “Invest, Then Buy My Chips”

This is Nvidia’s most direct investment logic. Nvidia injects capital into startups (like Poolside), who in turn use this money to purchase Nvidia’s expensive GB300 or next-generation GPUs. This not only locks in customers but also artificially creates demand, forming a “mutually beneficial” business ecosystem between Nvidia and its invested companies.

2. Strategic Objective: Building a “CUDA Ecosystem Moat”

Nvidia’s true moat is its CUDA software platform. By scattering money across all levels of the AI ecosystem (robotics, biopharmaceuticals, autonomous driving, AI programming), Nvidia ensures that these most promising future companies develop from day one based on the CUDA platform. This prevents competitors (like AMD, Intel) from shaking Nvidia’s position even if their hardware keeps up in terms of performance – they simply can’t disrupt Nvidia’s dominance in the software ecosystem.

3. Expansion Objective: Securing New Tracks for Chips

Nvidia needs to constantly find the next “AI wave” to support its $500 billion valuation. Its investment in Nokia (6G) is to push its AI chips into the massive telecommunications infrastructure market; its investment in Figure AI (robotics) is a bet on “embodied intelligence” becoming the next major consumer of computing power after cloud computing.

Summary: “Trickery” Under Monopoly

The unprecedented concentration in the US stock market reflects the productivity leap brought about by the AI revolution, but it also carries enormous risks of bubbles. The billion-dollar “closed loop” partnership between Nvidia and OpenAI is a microcosm of this high-stakes gamble.

Nvidia’s investment strategy is a meticulously planned “trickery”: It leverages its strong capital strength to not only become a “weapons supplier” in the AI era, but also to transform into a “banker” and “rule maker,” ensuring that all paths to the future are paved with Nvidia chips. This strategy is highly rational and efficient from a business perspective, but it also exposes it to significant antitrust pressure and potential financial risks.

For investors, understanding this game driven by AI, led by giants, and strengthened by capital closed loops is key to comprehending the current market.

From Meituan’s Losses to Bond Mismatch

Thu, 18 Sep 2025 19:19:10 +0800

In the turbulent and unpredictable stock market, we often use faith and expectations as our compass, attempting to navigate through the fog and reach the shores of wealth. However, when our voyage deviates from the guidance of the compass, it’s easy to lose direction, even to run aground. This essay is about one such journey, beginning with an unwavering obsession with Xiaomi, which rose and fell repeatedly amidst the waves of capital.

The Clash of Faith and Reality: Xiaomi’s Gains and Losses

In June, I bought into Xiaomi with unwavering faith in Lei Jun and a hopeful vision for the company’s electric vehicles. At the time, faith was a powerful force, temporarily blinding me to Lei Jun’s shrewdness as a capitalist and overlooking the risks posed by the High-Level Flash Lightning Distribution Platform. Fortunately, several wave trading operations ended with modest profits, giving me a sweet taste of entering the Hong Kong stock market for the first time.

However, this self-confidence soon followed. After making a small profit on Jiu Fang stocks, I became somewhat dazed, relying solely on the flow of net buying funds from the Hong Kong Connect indicator instead of rational analysis.

Guided by the flow of funds, I turned my attention to Meituan. However, this time, I didn’t delve into the company’s financial data, its historical trends, or even realize the raging food delivery war that was unfolding. When the stock price began to fall, I failed to cut my losses in time and instead blindly added to my position.

Shares trading at HK$20,000-30,000, limitations on capital prevented me from managing my positions freely, which undoubtedly exacerbated my losses in Meituan. This lesson deeply made me realize that in the Hong Kong stock market, blind re-positioning is like trying to fill a constantly leaking bathtub with more water – it only accelerates the loss.

“Stupid People Are Lucky” - An Unexpected Gain

While I was worrying about Meituan’s losses, an unexpected turn of events occurred – I bought Alibaba. Although I didn’t hold onto it, I quickly sold it and used the profit to offset some of Meituan’s losses. This might be what’s meant by “stupid people are lucky,” allowing me to temporarily escape Meituan’s predicament.

Finance: More Important Fundamentals Than Investing

This experience also made me re-examine my financial management approach, aside from the ups and downs of the stock market. At the time, I invested a sum intended for short-term emergency funds into Chinese bonds. This money was meant to be used for payment within two months, but the short-term holding yield on the bonds wasn’t high, and it also posed liquidity issues. If I had instead chosen to deposit this money into a money fund, I could have earned more stable returns and readily redeemed it, better meeting my funding needs.

AI Investment on Uncle Xiang's Notebook

In a low interest rate environment, I have rearranged a ten-year financial plan.

From yield changes back to the cash flow question

Real-world Constraints of Two Cross-Border Configurations

Reviewing History with a Set of Reproducible Standards

Backtesting Didn’t Make the Decision for Me, But It Changed How I Frame the Question

Data Description

References

Original Prompt

Deregulation and Funding News Appear on the Same Day — How Zhipu and MiniMax Are Being Repriced

First, Look at the Price Path

The Price of New Money, and the Price of Old Money

Implied equity value with similar IPO timing, later placed into different deal structures

There is still one more round for A-shares, but the amount cannot be factored in in advance

LLM Funding Burns On — What Is the Market Watching Now?

References

Original Prompt

The AI spending bull arrives at the lockup expiration

Programming First Becomes a Cash Flow Entry Point

Where Zhipu’s Market Value Comes From

What Really Changed After the July Lift

Why Have Many A-Shares and Hong Kong Stocks Not Risen

References

Original Prompt

Writing Approach Summary

Xiaomi drops back to hardware ledger: Storage price hikes, auto subsidies, and AI spending hit simultaneously

Stock Price First Fell Due to Profit Doubts

Several Headwinds Are Not the Same Kind of Pressure

How to Buy a Phone Amid Rising Storage Costs

Focus on the Substance, Not the Slogans

References

Original Prompt

Summary of Conceptual Ideas

After an AI stock pullback, first look at how the crowded trades unravel/dissipate.

A-Shares Continue to Decongest

Mid-day Rebound in US Stocks Does Not Signal Cleansing Completion

See three lines below

References

Original Prompt

This Update’s Prompts

Summary of Writing Ideas

The renewed surge of A-share semiconductors should not be bought based on industrial logic alone; caution is needed as this may constitute a crowded trade.

Let’s first see how much it has risen in this round

It’s Not Just AI: Many Concepts of “True Logic” Have Gone Through Their Failures

With this batch of companies now, are their earnings supporting the stock price?

Increasing Semiconductor Positions Now: The issue isn’t whether it is right or wrong, but the timing (position).

References

Original Prompt

Zhipu and MiniMax entering Hengke; rules and buying pressure are two completely different things.

I cannot see the “Unlisting” item in Hengke’s rules.

What Exactly Changed in This Announcement

De-listing and Index Inclusion Are Not the Same Thing

Inclusion Does Not Guarantee Gains

My Perspective on Zhipu and MiniMax

References

Writing Notes

Original Prompts

Writing Idea Summary

U.S. Treasury Yield at 4.5%: What Will I Actually Get After Buying?

Conclusion First

What is the 4.5% actually?

Calculating with an Official New Bond

Why is the coupon rate 4.375%, but the yield to maturity is 4.468%?

Whether the Yield Can Truly Be Understood as 4.5% if It Never Sells

First, 4.5% is more like an annualized rate, not a cash payout rhythm/disbursement schedule

Secondly, the money in the account will not automatically compound 4.5%

When Will It Be Issued/Distributed

Treasury Bills

Treasury Notes and Treasury Bonds

I think the most important thing to clarify first is not the yield rate, but the definitions/scope

References

Writing Notes

Original Prompt

Writing Outline Summary

Extended Brainstorming

Why didn't the Hang Seng Tech index hit new highs alongside A-shares?

Clarifying the Scope

Why A-Shares Are Strong While Hang Seng Tech Is Pulling Back

Are Hang Seng Tech and Chinese Concept Stocks Strongly Correlated?

China Concept Gold Dragon and NASDAQ: Is there a strong correlation?

What is the `4.5%` actually?

Whether the Yield Can Truly Be Understood as `4.5%` if It Never Sells

First, `4.5%` is more like an annualized rate, not a cash payout rhythm/disbursement schedule

Secondly, the money in the account will not automatically compound `4.5%`

`Treasury Bills`

`Treasury Notes` and `Treasury Bonds`