Finance
“China's AI bubble has burst, wiping out ¥3.2 trillion in just four months.”
Plain restatementChina's AI-heavy STAR50 index has fallen sharply from its July 2026 peak, and the decline corresponds to roughly ¥3.2 trillion of lost market value over a period of about four months.
General information only - not financial advice.
This post is mostly accurate and closely tracks its source. Reuters reported on 26 October 2026 that China's STAR50 index of AI and semiconductor companies had fallen 40% from its July peak in under four months, wiping out about $480 billion in market value, and Reuters itself used the "bubble has already burst" framing in its headline. The ¥3.2 trillion figure in the post is a correct conversion of that $480 billion at the exchange rate on the day Reuters published, though no source actually states the yuan number. Two smaller points are worth knowing. The tally covers one 50-company index rather than every AI-linked Chinese stock, and the separate "1.3 trillion erased today" line inside the post's images comes from a social media account, not from Reuters, and carries no currency unit. What remains uncertain is the precise size of the loss, since the $480 billion is Reuters' own calculation rather than a figure published by the exchange, and whether a 40% fall amounts to a burst bubble is an interpretation rather than a measurement. The decline was still continuing at the time of this check, with the index falling a further 4.8% on 28 October 2026 to its lowest level since March.
China's [drifted from the evidence:] AI bubble has [drifted from the evidence:] burst, wiping out ¥3.2 trillion [drifted from the evidence:] in just four months.
China's [added by the neutral restatement:] AI-heavy STAR50 index has [added by the neutral restatement:] fallen sharply from its July 2026 peak, and the decline corresponds to roughly ¥3.2 trillion [added by the neutral restatement:] of lost market value over a period of about four months.
Red-tinted words in the claim drifted from the evidence. Green-tinted words are what a neutral restatement needs.
The trace / claim to source
- The ¥3.2 trillion figure is arithmetically correct. It is the dollar value Reuters reported, converted at the USD/CNY rate prevailing on the day Reuters published, 26 October 2026.
- The 40% decline is as Reuters reported it, measured on the STAR50 index from its July 2026 peak.
- "Four months" matches the source closely. Reuters wrote "less than four months."
- The "bubble has burst" framing is not the poster's invention. It is the wording of Reuters' own headline and the thrust of its analysis.
- The individual stock figures in the post track the source. Unitree halving, Moore Threads and MetaX each losing about half their value since July, and optical component makers down around 60%, all correspond to what Reuters reported for those named companies.
- The valuation comparison is accurate as reported: the STAR50 near 100 times trailing earnings against roughly 30 times for the Nasdaq 100, as of the Reuters analysis on 26 October 2026.
- The 42% of market capitalisation and 57% of turnover figures are correctly reproduced, and correctly belong to Goldman Sachs, as cited by Reuters. The scope is A-shares plus Hong Kong H-shares combined, not mainland shares alone.
- The Nasdaq comparison is accurate as of the Reuters analysis: roughly 6% below its record high.
- The claim has not gone stale in the direction that would undercut it. The STAR50 fell a further 4.8% on 28 October 2026 to its lowest level since March, so the loss as of the verdict date is larger than the figure quoted, not smaller.
- Scope stretch on the headline figure: the claim's caption says "China's AI stock market has reportedly lost ¥3.2 trillion," but the tally Reuters produced is the market value lost by the STAR50 index specifically, a 50-company index. Chinese AI-linked equities extend well beyond those 50 names, including Hong Kong-listed companies. The figure is a floor for one index, presented as a total for a sector. This widens the apparent scope without changing the order of magnitude.
- Unattributed currency conversion: the post presents ¥3.2 trillion as though it were the reported figure. No source states ¥3.2 trillion. Reuters stated $480 billion. The conversion is correct, but a reader searching for the yuan figure will not find it in the record, and the conversion silently inherits whatever rounding sits inside Reuters' own approximation of "some $480 billion."
- A characterisation presented as a measurement: "the bubble has burst" is an interpretive judgment about what a price decline means, not an observed quantity. Reuters does assert it in its headline, and separately quotes Beijing fund manager Dong Baozhen describing a speculative bull turning into a bursting bubble. The 40% and the $480 billion are measurements. "Burst" is not, and a reader cannot check it the way they can check the index level.
- Unsourced figure inside the image: the slide reading "1.3 TRILLION erased from the Chinese stock market today" is not from Reuters. It traces to a social media account, states no currency, and is reproduced elsewhere as ¥1.43 trillion. It sits beside correctly sourced Reuters numbers, which lends it credibility the sourcing does not support.
- The $480 billion loss is Reuters' own tally rather than an exchange-published figure. I could not reproduce it from a Shanghai Stock Exchange or index-provider market capitalisation series, so the precise magnitude rests on Reuters' calculation method, which the article does not set out.
- Every figure in the claim traces to one Reuters analysis published 26 October 2026. The dozens of outlets carrying it are syndication of that single piece, not independent confirmation. The underlying index decline is independently observable in market data; the dollar tally is not.
- Whether the decline constitutes a "burst bubble" as opposed to a severe drawdown that partially retraces is not resolvable from price data at this date. The STAR50 was still falling on 28 October 2026.
- The exact composition of the "AI stocks" basket behind Goldman Sachs's 42% and 57% figures is not specified in the available text, so the sensitivity of those percentages to classification choices cannot be assessed.
Reuters reported on 26 October 2026 that the mainland STAR50 index of AI and semiconductor leaders had fallen 40% in less than four months, vaporising some $480 billion in market value. Reuters cited Goldman Sachs for AI stocks running to 42% of China's market capitalisation and 57% of turnover across mainland-listed A-shares and Hong Kong's H-shares. On individual names, Reuters reported that shares in robot-maker Unitree, 8,000 times oversubscribed by retail buyers at listing, had halved since they started trading in August, and that chipmaker CXMT had lost more than a quarter of its value in the past month. On the optical component makers, Reuters reported that shares in Zhongji Innolight and Eoptolink Technology were down roughly 60% from highs in June and July, in part as Washington moved towards banning them from supplying US data centres. On valuation, Reuters put the STAR50 near 100 times trailing earnings, compared with nearly 250 times at its earlier high, and contrasted that with roughly 30 times for the Nasdaq 100. On the US comparison, Reuters was reported as noting that the Nasdaq sat roughly 6% below the all-time record high it established three weeks earlier. On the broader market, the CSI300 index had also fallen about 20% from its June high. The selloff was still running at the verdict date. Chinese stocks closed at a 14-month low on Tuesday 28 October 2026, with the Shanghai Composite down 1.3% at 3,648.30, the lowest since August 2025, and Shanghai's tech-focused STAR50 index down 4.8% to its lowest since March. Currency arithmetic: USD/CNY closed at 6.6703 on 26 October 2026. $480 billion multiplied by 6.6703 equals ¥3.202 trillion. The claim's ¥3.2 trillion is therefore a correct conversion of the Reuters dollar figure at the prevailing rate on the date Reuters published.
Complete reasoning
The reply is formatted for pasting into the thread where the claim is circulating.
Ask this case
Answers come only from the case file above; nothing is added.
Did China's AI stock market really lose ¥3.2 trillion?
The ¥3.2 trillion figure is a correct conversion of $480 billion, the amount Reuters reported was wiped off China's STAR50 index over about four months. No source actually published a yuan figure; it was calculated using the exchange rate on the day Reuters reported the story.
Does the ¥3.2 trillion cover all Chinese AI stocks?
No. It reflects losses in the STAR50, a 50-company index of AI and semiconductor firms, not every AI-linked Chinese stock. Many AI-related companies, including some listed in Hong Kong, sit outside that index, so the figure understates the sector's full scope while being presented as if it covers it.
Is it accurate to call this an AI bubble that has burst?
That phrase comes from Reuters' own headline and reflects its analysis, so the post did not invent it. But it is an interpretation of what the price decline means, not a measurable fact like the 40% drop or the dollar figure, and cannot be checked the same way.
What about the separate claim of 1.3 trillion lost in a single day?
That figure appears in the post's images but does not come from Reuters. It traces to a social media account, lacks a stated currency, and is reported elsewhere as ¥1.43 trillion, so its accuracy is not established by the case file.
Has the decline gotten better or worse since the ¥3.2 trillion figure was reported?
Worse. The STAR50 fell another 4.8% on 28 October 2026 to its lowest level since March, meaning the actual loss by that date was larger than the $480 billion Reuters originally reported.