Direct answer: yes, in China the AI bubble has already burst — but no, it did not happen in the last five days. Chinese equities have lost roughly $2.4 trillion in market value since May 2026, and the Shanghai STAR 50 index of AI and semiconductor leaders has fallen about 40% since June, wiping out roughly $480 billion on its own. What exploded into headlines in late October was the verdict, not the event: a five-month rout that turned a two-year AI frenzy into a bear market. Reuters put it bluntly in its analysis this week: in China, the AI bubble has already burst.
The confusion is understandable. Social media posts this week compressed months of losses into a single dramatic weekend, and the real numbers are dramatic enough without compression. This article separates the two figures people keep mixing up, explains how the bubble inflated, what finally popped it, and why the American market — where the same AI debate rages — has not followed China down.
The numbers: what actually vanished
Two headline figures are circulating, and they measure different things over different windows. Adding them together, or treating one as the other, is the most common error in this story.
| Market or index | Fall | Window | Value wiped / note |
|---|---|---|---|
| All Chinese equities | Bear market | Since May 2026 peak | Roughly $2.4 trillion in market value |
| Shanghai STAR 50 (AI and chip leaders) | About 40% | Since June / July 2026 | Roughly $480 billion |
| CSI 300 (blue-chip index) | About 20% | From its June peak | Bear market; 14-month low in late October |
| Nasdaq (US, for comparison) | About 6% | From its record high | A pullback, not a burst |
The $2.4 trillion figure counts the drop in mainland share-market value from the May peak, according to Bloomberg's October 22 bear-market reporting. The $480 billion figure is much narrower: it covers only the STAR 50's AI and chip stocks, per Reuters' analysis of October 26. The CSI 300's bear-market maths runs from its June high, a different starting point again. Same storm, three different rulers — which is why honest reporting keeps the three figures apart.
How the bubble inflated
The rally was not a fringe mania. At its peak, AI-linked stocks accounted for roughly 42% of the total market capitalisation of mainland China and Hong Kong combined, and roughly 57% of trading volume, according to Goldman Sachs figures cited in the October coverage. When nearly half the market is one trade, that trade's fate is the market's fate.
Three forces fed it. First, state backing: Beijing has made AI and technological self-reliance a national priority, and state-linked money poured into chipmakers, robotics firms and data-centre builders. Second, genuine excitement: Chinese AI models and humanoid-robot demos convinced investors that the country could close the gap with the United States. Third, a retail frenzy: debuts surged — GPU makers Moore Threads and MetaX jumped roughly 400% and 700% on their market debuts — and each listing success pulled more money into the next one.
The frenzy had a founding moment. In January 2025, Chinese startup DeepSeek released a chatbot that matched leading Western models at a fraction of the reported training cost, and the shock briefly erased about $1 trillion from US stocks in a single day. For Chinese investors, the message was intoxicating: the gap was closable, and it might close fast. Money that had been waiting for a reason to believe poured in — and stayed in, through two years of listings, demos and policy speeches, until the earnings stopped cooperating.
Valuations lost contact with earnings. At the peak, STAR 50 companies traded at nearly 250 times their past earnings. For comparison, the Nasdaq 100, the US tech benchmark, trades at roughly 30 times. Paying 250 years of current profits for a share only makes sense if profits explode soon. They did not.
The stocks that fell hardest
The damage was worst exactly where the euphoria had been hottest. Humanoid-robotics star Unitree has roughly halved in value since its August listing. GPU makers Moore Threads and MetaX, the debut-day rockets, have each lost roughly half their peak value since July. AI firms MiniMax and Zhipu have reportedly fallen more than 85% from their peaks — a figure carried as reported, not independently confirmed, but consistent with the scale of the washout at the speculative edge.
The pattern matters: this was not one bad company or one bad sector. Robotics, chips, models and data-centre suppliers fell together, which is what a bursting bubble looks like as opposed to a normal correction. When everything bought for the same story falls at once, the story itself was the trade.
What popped it
No single pin did it. Four pressures converged over the summer and autumn.
Profit doubts. The central question — when does AI spending turn into profit? — kept getting the wrong answer. Dan Wang, China director at Eurasia Group, captured the mood in the October coverage: profit margins fell short of expectations, and it did not look as though embodied intelligence, the robotics bet, could turn into mass-market profits any time soon.
Beijing's own warning. On October 9, Beijing issued guidelines aimed at preventing investment bubbles in "new productive forces", a category that includes AI. Markets read it as a signal: the state that had cheered the rally on was now worried about its size. For a rally built partly on state backing, that was a cold draught.
Weak domestic demand. China's consumers never joined the party. Households stayed reluctant to spend, and the tech boom did not lift their incomes. A market priced for an economic transformation kept meeting an economy where shoppers were cautious.
Western trade curbs. US restrictions kept tightening around the edges of China's AI build-out — including, in July 2026, US Federal Communications Commission restrictions on newly released foreign-made humanoid and quadruped robots, plus connected power inverters and data-centre infrastructure. Each curb raised the cost and the risk of the whole AI supply chain.
Layered on top, higher US interest rates made speculative bets everywhere more expensive to hold. None of these alone bursts a bubble. Together, they removed every reason to keep paying 250 times earnings.
Even the state's own market support looked thinner. Bloomberg Intelligence analysts estimated that Central Huijin, the state-backed investor that leads China's market stabilisation, raised about $231 billion through ETF sales in the first half of the year and had put back only about $43 billion since buying resumed in July. The "national team" had not abandoned the market — but it was no longer cushioning every fall, and traders noticed.
Why America did not follow
The obvious question: if the AI bubble burst in China, why is the Nasdaq only about 6% off its record high? The short answer is that the two rallies were never the same trade.
The valuation gap tells much of the story. Even after its 40% fall, the STAR 50 still trades at roughly 100 times past earnings — more than triple the Nasdaq 100's roughly 30 times. China's rally was broader, more retail-driven and more dependent on policy support, with thinner profits underneath. America's AI leaders, whatever one thinks of their valuations, report enormous profits from products already selling. The US debate is about whether those profits justify current prices; China's debate became whether the profits would arrive at all.
That does not make American investors safe. It makes the comparison useful: a 6% pullback is a market having doubts, while a 40% collapse in the flagship AI index is a market that has already rendered its verdict. Reuters' line — in China, the bubble has already burst — is a statement about China, not a prophecy about everywhere else.
Contagion is still worth watching, through two channels. Global funds that hold both markets may sell US winners to cover Chinese losses, and a weaker Chinese tech sector buys fewer components and less equipment from suppliers in South Korea, Japan and Taiwan — markets that already fell in sympathy on the worst October days. Neither channel has so far turned the American pullback into a rout, but both are reasons the US market is unlikely to ignore China entirely.
What happens next: what to watch
Morgan Stanley has reportedly warned that Chinese stocks could fall another 10% to 15% before finding a bottom. Treat that as a reported warning, not a forecast to trade on — but it signals that professional investors do not yet think the washout is finished.
Four things will decide whether this is the end of the rout or only the middle of it. First, earnings: do AI firms start converting spending into profit, or do margins keep disappointing? Second, Beijing: do the October 9 guidelines stay a warning, or do they harden into tighter controls on AI investment? Third, valuations: the STAR 50 at roughly 100 times earnings is cheaper than 250 times, but still expensive by any global standard — further falls are possible without any new bad news. Fourth, domestic demand: a consumer recovery would give the whole market a floor that speculation alone cannot.
One more caution for readers of viral posts: market value lost is not cash burned. The $2.4 trillion is the fall in what investors collectively think shares are worth, not money the companies spent or revenue they lost. It still matters — pensions, savings and state funds hold these shares — but it is a repricing, not a vault emptied. And the timeline matters most of all: this was a five-month unwinding, confirmed in late October, not a five-day crash. Anyone telling you otherwise is selling speed, not accuracy.
None of this is investment advice, and this article makes no call to buy or sell anything. For ordinary readers, the practical takeaway is simpler: bubbles rarely announce themselves at the top, they announce themselves in hindsight — in a valuation multiple nobody questioned, a debut pop nobody examined, and a warning from the very government that cheered the rally on. Those were all visible before October. They are worth remembering the next time a market, anywhere, starts telling you that this time the profits can wait.
Frequently asked questions
Did China's AI bubble really burst in the last five days?
No. The losses accumulated from May to October 2026 — roughly $2.4 trillion across Chinese equities, with the STAR 50 down about 40% since June. Late-October coverage confirmed the burst; it did not report a five-day event.
Are the $2.4 trillion and $480 billion figures the same loss counted twice?
No. The $2.4 trillion figure covers all mainland Chinese equities since the May peak (Bloomberg, October 22). The $480 billion figure covers only the STAR 50's AI and chip stocks (Reuters, October 26). They must not be added together.
Does this mean AI itself has failed in China?
No. It means the share prices ran far ahead of the profits. Chinese AI research and products continue; what burst was the valuation — paying nearly 250 times earnings for companies whose profits had not arrived.
Is the STAR 50 cheap now after falling 40%?
Not by global standards. At roughly 100 times past earnings it is down from nearly 250 times, but still more than triple the Nasdaq 100's roughly 30 times. Cheaper is not the same as cheap.
Could the same thing happen to US AI stocks?
The US market is only about 6% off its record high, and its AI leaders report large existing profits — a different starting point. That does not rule out a US correction, but the Chinese burst is not, on its own, evidence that one is imminent.
Sources
- Bloomberg, October 22, 2026: China stocks set for bear market as AI trade and policy hopes fade — https://www.bloomberg.com/news/articles/2026-10-22/china-stocks-set-for-bear-market-as-ai-trade-policy-hopes-fade
- Reuters analysis, October 26, 2026: "In China, the AI bubble has already burst" (syndicated by Mint) — https://www.livemint.com/ai/chinas-ai-stock-bubble-bursts-wiping-out-480-billion-as-chipmakers-plunge-up-to-60-amid-valuation-fears-11793109217265.html
- Goldman Sachs figures (AI share of market capitalisation and trading volume) as cited in the October 2026 coverage above
- Corroborating late-October coverage of the same figures — https://theenglishchronicle.com/News/33926/ · https://www.tradingview.com/news/coinpedia:6bc090452094b:0-china-stock-market-crashes-has-the-ai-bubble-finally-burst/ · https://demo.decrypt.co/381992/china-stock-market-lost-2-4-trillion-investors-doubt-ai-boom
Be the first to share a useful question or thought.