Chen Wensheng runs a small business in Hainan, and like millions of ordinary Chinese savers, he put money into the AI rally this year. He told Reuters he felt emboldened because Beijing seemed to be backing names like robot maker Unitree, which counted state-backed funds among its investors. That feeling of official blessing is gone now. He and his friends are nursing losses.
Their story is the human side of a number that has been all over my feed this week: about $480 billion. That is how much market value disappeared from China's AI and semiconductor stocks in less than four months. When I first saw the headline I assumed it was exaggerated. It isn't. If anything, the real story is more interesting than the headline, because the crash didn't come from one dramatic event. It came from a lot of small cracks that finally gave way together.
Here's what happened, why it happened, and what I think it means.
What actually happened to China's AI stocks
The STAR50, a Shanghai index packed with Chinese AI and semiconductor leaders, has dropped nearly 40% since July. The wider CSI 300, which tracks China's biggest listed companies, is down about 20% from its June high and has slipped into bear-market territory. Meanwhile, the Nasdaq, home of the US AI trade, is only about 6% off its record.
| Measure | Figure | Source |
|---|---|---|
| STAR50 fall from July peak | About 40% in under four months | Reuters |
| Market value erased | Roughly $480 billion | Reuters |
| CSI 300 fall from June high | About 20% | Reuters |
| AI share of China's market value | 42% | Goldman Sachs |
| AI share of A-share and H-share turnover | 57% | Goldman Sachs |
| STAR50 P/E ratio, now vs peak | About 100x vs nearly 250x | Reuters |
| Nasdaq 100 P/E ratio | About 30x | Reuters |
| Nasdaq drop from record | About 6% | Reuters |
One detail deserves a second look. Even after losing 40%, the STAR50 still trades at roughly 100 times trailing earnings, more than three times the Nasdaq 100's multiple. In plain words: after the crash, Chinese AI stocks are still expensive. That is exactly why several analysts worry the fall isn't finished.
How the bubble got built
Bubbles never appear overnight, and this one had a clear two-year runway. Stimulus promises from Beijing pushed Chinese shares up and kept them rising. Then AI became the main story. Hong Kong listings of model developers Zhipu (now trading as Z.AI) and MiniMax in January gave investors a way to bet directly on Chinese large language models. By late February, Reuters Breakingviews was already pointing out that the two had a combined value near $63 billion after more than quadrupling.
The IPO mania was the loudest signal. Retail buyers oversubscribed Unitree's share offering about 8,000 times. Chipmakers MetaX and Moore Threads jumped roughly 700% and 400% on their debuts last year. By the time the rally peaked, Goldman Sachs estimates that AI-linked stocks made up 42% of China's total market value and 57% of trading turnover across mainland and Hong Kong shares. That's not a sector anymore. That's the whole market leaning one way.
Some people saw it coming. In June, Chinese hedge fund Wealspring Asset called AI a "super bubble" and warned the hottest infrastructure names could fall more than 80%, comparing the mood to China's 2015 frenzy. They were early, but they weren't wrong.
Why China's AI bubble burst: the real reasons
1. Valuations left no room for error
At the peak, the STAR50 traded near 250 times earnings. According to analysis by 36Kr reported by Digit, Zhipu was valued at around 150 times annualised revenue in early October, with MiniMax at roughly 40 times. When you pay prices like that, the company has to deliver perfection for years. Any hint of slower growth and the whole story wobbles.
2. Small investors drove the rally, and small investors run for the exit
With AI stocks taking up such a big share of turnover, few ordinary investors escaped. Beijing had actually been trying to calm things down. Regulators cracked down on leverage and high-frequency trading, and state-backed investors sold into rallies to cap gains. On October 9, authorities even issued guidelines aimed at preventing bubbles in so-called "new productive forces," a category that includes AI. The crowd ran straight past all of it. I find this the most ironic part of the whole saga: a government that wanted a slow, boring bull market got a stampede instead.
3. Free models destroyed pricing power
China's AI industry has been in a price war since 2024, when DeepSeek slashed prices and ByteDance, Alibaba and Baidu followed. This year the pressure only grew. When DeepSeek previewed an open-source V4 model in April, Zhipu and MiniMax shares each dropped about 9% in a single session. Moonshot AI's open-source Kimi K3 in July kept the heat on. If a good model is free, charging for yours gets hard. Giants like Alibaba, ByteDance and Tencent can fund losses from other businesses. Zhipu and MiniMax cannot, and MiniMax also leaned on consumer apps for about two-thirds of its 2025 revenue.
4. Export controls and the threat of overseas bans
Traders told Reuters that fear of the US or other countries banning Chinese equipment from data centres played a role. Optical component makers Zhongji Innolight and Eoptolink are down about 60% from their June and July highs, partly because Washington is moving toward barring them from supplying American data centres. On the chip side, IDC data reported by Reuters shows Nvidia's share of China's data-centre AI accelerator market fell to 34% in the first half of 2026, with Huawei filling the gap. Self-reliance sounds great, but it comes with an upfront bill.
5. Capacity got built faster than demand
China built data centres at speed. As far back as 2025, reports (including DigiTimes coverage cited by TechRadar) described hundreds of projects launched across 2023 and 2024 with actual usage lagging, and some operators selling off GPUs. Investors have been asking the same awkward question their American counterparts ask: where exactly is the revenue that justifies all this spending?
6. The money left, at home and abroad
Bloomberg Intelligence analysts say the global AI hardware selloff hit China's A-shares hardest, because foreign investors had piled into the sector in the second quarter. Outflows through the Hong Kong trading link reached an estimated 90 billion yuan (around $13.5 billion) in the third quarter, the biggest in a year. Add tax bills and nervous retail holders and you get a market with no buyers.
Which companies took the biggest hit
| Company | What it does | What happened |
|---|---|---|
| Unitree | Robot maker | Roughly halved since its August listing, after retail oversubscription of about 8,000 times |
| MetaX Integrated Circuits | Chipmaker | Up about 700% on debut last year, now halved since July |
| Moore Threads | Chipmaker | Up about 400% on debut last year, now halved since July |
| CXMT | Memory chipmaker | Down more than a quarter in the past month |
| Zhongji Innolight, Eoptolink | Optical components | Down about 60% from June and July highs |
| Zhipu (Z.AI) and MiniMax | AI model developers | Combined $150 billion below their highs, per Bloomberg; MiniMax fell below its IPO price on October 22 |
The list reads like a who's who of last year's darlings. And the damage isn't only about portfolios. Fund manager Hong Hao of Lotus Asset Management told Reuters the rout is likely to stall many companies that were hoping to list, which means fewer funding options for the next wave of Chinese AI startups.
Burst, or just a violent correction?
Here's where I want to slow down, because the phrase "bubble burst" can mislead. The $480 billion is a loss of market value, not cash that vanished from company bank accounts. Analysts quoted by Coinpaper make that point clearly, though they add that lower valuations make it harder for startups to raise the equity they need for chips and computing.
There are also signs of life. Bloomberg reported that DeepSeek recently raised $15 billion in one of the largest funding rounds in Chinese history. Chinese labs keep shipping models. DRAM contract prices are still expected to rise 10% to 15% in the fourth quarter, according to TrendForce. And some big banks aren't running for the doors: Goldman Sachs still lists Innolight and CXMT among its recommended Chinese AI stocks, and BNP Paribas strategist William Bratton sees buying chances for careful investors.
On the other side, Morgan Stanley expects offshore Chinese stocks to fall another 10% to 15% before they find a floor. One Shanghai fund manager, Wang Zhuo, put it bluntly to Reuters: chipmakers and AI suppliers will end up in a price war like every other Chinese manufacturing sector, and in his words, "All bubbles inevitably burst."
My take: three things I keep coming back to
First, this is a story about people, not just prices. The loudest voices in a crash are fund managers and analysts, but the ones who lose the most are savers like Chen in Hainan who believed an official signal meant safety. Anyone who has watched India's own IPO frenzies knows the pattern: a hot listing, a crowd that treats a subscription number as proof of quality, and a long, quiet hangover.
Second, don't read this as "China's AI has failed." Share prices and technology are different things. The models are still improving and open-source releases from DeepSeek, Zhipu and Moonshot keep pushing the global frontier on price. What cracked is the idea that every AI company deserves a trillion-yuan story. In my view, that correction was overdue, and it may actually be healthy for the survivors.
Third, the West shouldn't relax. The Nasdaq is only about 6% down, but the spending numbers are enormous. Company forecasts cited by Digit put 2026 capital spending at roughly $180 to $190 billion for Alphabet, about $190 billion for Microsoft and around $200 billion for Amazon. China's crash is a preview of what happens when expectations outrun earnings. Whether the US follows is the real question, and honestly, nobody knows yet.
What it means for Indian readers
Indian markets have relatively few pure AI plays, so the direct impact is limited. Where it could matter is indirect: global and Asia-focused funds with heavy China exposure, and the broader mood around technology and AI-linked stocks. Indian IT services companies also face their own AI-disruption worries, which is a separate story. If you hold international or emerging-market funds, check how much China weighting sits inside them.
What to watch next
A few dates and pressures could decide whether the slide continues:
January lock-up expiry. Pre-IPO shareholders will be able to sell about 178 million Z.AI shares and 25 million MiniMax shares in January, a potential new wave of supply.
New listings. IPOs from DeepSeek and Moonshot AI could pull money away from the already-listed labs.
US earnings and policy. Microsoft, Alphabet and Amazon were due to report this week, which is the next test of the global AI trade, and any US move to ban Chinese optical suppliers would add pressure.
Beijing's response. One analysis argues Beijing is unlikely to bail out the AI trade, preferring to protect strategic areas like chips and computing power while shareholders and local governments absorb the losses.
The bottom line
China's AI bubble burst for a simple reason with a messy backstory: people paid tomorrow's prices for today's earnings, and then the story around demand, competition and politics got complicated all at once. The stock market has taken a brutal reset. The technology, the talent and the ambition are still there. The next few months, especially January's share unlocks and the fate of the big AI IPOs, will show whether this was a painful clean-out or the first leg of something longer.
Frequently asked questions
Did China's AI bubble really burst?
In the stock market, yes. The STAR50 index of Chinese AI and chip companies fell about 40% from its July peak in under four months, erasing roughly $480 billion, and Reuters reported that the bubble has already burst. The underlying technology is not collapsing, though. Chinese labs are still releasing models and raising money.
Why did China's AI bubble burst?
Prices ran far ahead of profits. Valuations on the STAR50 peaked near 250 times earnings, retail investors piled in, free open-source models and price wars crushed pricing power, and fears of export curbs and bans on Chinese equipment in overseas data centres spooked buyers. Reuters said there was no single trigger.
How much money did China's AI stocks lose?
About $480 billion in market value, according to Reuters. That is paper value, not cash that left the companies. The broader CSI 300 index is also down roughly 20% from its June high.
Which Chinese AI companies were hit hardest?
Robot maker Unitree has roughly halved since its August listing. Chipmakers MetaX and Moore Threads have halved since July. Optical component makers Innolight and Eoptolink are down about 60% from their highs, and model developers Zhipu (Z.AI) and MiniMax have lost a combined $150 billion from their peaks, according to Bloomberg.
Does the crash mean China's AI technology has failed?
No. A falling share price is not the same as failing technology. DeepSeek reportedly raised $15 billion recently, and Chinese labs continue to ship open-source models. What has cracked is the valuation story, not the research.
Will China's AI crash spread to the US and global markets?
It could add pressure, but the damage so far is uneven. The Nasdaq is about 6% below its record, far less than the STAR50 drop. Investors are watching US big tech earnings and AI spending closely for signs the worry is spreading.
Is it a good time to buy Chinese AI stocks now?
Opinions are split, and this is not financial advice. Goldman Sachs still recommends some names and BNP Paribas sees opportunities for careful buyers, while Morgan Stanley expects offshore Chinese stocks to fall another 10% to 15% before bottoming. The STAR50 still trades near 100 times earnings, so it is hardly cheap.
Sources and further reading
- Reuters (via Daily Sabah): China's stock plunge signals AI bubble may be bursting, October 27, 2026
- Reuters (via The Kathmandu Post): In China, the AI bubble has already burst
- Digit: China AI Bubble Burst explained, October 28, 2026
- Bloomberg: China AI Model Stocks' $150 Billion Drop Shows No Sign of Easing, October 23, 2026
- Futurism: It Appears That the AI Bubble Has Just Burst in China
- BeInCrypto (via Yahoo Finance): AI Bubble Bursts in China, Wiping $480 Billion
- Coinpaper: China's AI Stocks Just Lost $480 Billion
- Crypto Briefing: China's AI stock bubble has burst, Reuters reports
- TechRadar: Data centers in China are dumping rare 48GB Nvidia RTX 4090D GPUs
Disclaimer: This article is for information and news purposes only and is not financial or investment advice. Market figures are based on reports published up to October 29, 2026 and may change quickly. Always do your own research or speak to a licensed adviser before investing.