The China AI stock bubble has erased roughly $480 billion in market value, with the STAR50 index down about 40% from its July peak in under four months. The STAR50 is home to the country’s hottest AI and chip names. The figure comes from a Reuters report published Oct. 26 and cited by outlets worldwide.
The scale is easier to grasp with one number. Goldman Sachs says AI stocks make up 42% of market capitalization and 57% of turnover across mainland China and Hong Kong. That is a huge bet on a single theme, so a drop like this hurts almost everyone.
For American investors, this is not a faraway problem. If your 401(k) or index fund holds big U.S. tech names, you have a stake in how this story plays out. A 401(k) with a standard target-date or S&P 500 fund has little direct China AI exposure. But those same funds are loaded with the American AI leaders that investors now compare to China’s. When one AI market cracks, traders start asking about the other.
The Nasdaq is down about 6% from a record set three weeks ago. That is nowhere near China’s pain, but the timing is awkward.
The damage reaches well beyond one index. The broader CSI300 is down about 20% from its June peak. Geo News reports it has slipped into bear-market territory after ten straight weekly declines. The $480 billion is Reuters’ own tally, and it moves with prices every day.
The poster child is Unitree, the humanoid-robot maker. It listed on Aug. 19 at 150.80 yuan and closed its first day up 460%. Its first-day high of 1,100 yuan implied a valuation of about $66 billion, The Robot Report noted. Now the stock has roughly halved from that peak. It still trades far above its IPO price, which tells you how wild the ride was.
Chipmakers MetaX and Moore Threads have each lost about half their value. Memory chipmaker CXMT is down more than 25% in the past month.
The demand side looked very different not long ago. Chinese giants had ordered more Nvidia cards than the company can build, which shows how hot the buildout ran before this China tech stock rout.
Why did the Chinese AI stocks rout happen?
Nobody can point to one trigger. Reuters says traders cited lofty valuations and tax bills. They also pointed to the possibility of U.S. or other bans on Chinese equipment in data centers.
The valuation math is hard to ignore. The STAR50 still trades at about 100 times trailing earnings, Finimize reports. That is down from nearly 250 times at the peak. It is still more than three times the roughly 30 times the Nasdaq 100 trades at.
A Beijing fund manager told Reuters, “And now we see the bubble bursting.” That is a fragment of a longer quote, so take it as one voice, not a verdict.
Not everyone agrees the bubble has actually burst. Some skeptics say a 40% drop from a 250-times peak is a reset, not a collapse.
Who is holding the bag in the China AI bubble?
Mostly regular people. STAR50 holders are largely small retail investors. That matters because Beijing has been pushing household savings into stocks to help fund the AI buildout. A two-year rally has now run into a wall, and that plan looks a lot shakier.
The government is not sitting still. Inflows into nine ETFs actively traded by state-backed investors hit 128.7 billion yuan, about $19 billion, last week. That is the most since April 2025, Bloomberg data shows. The selloff has also raised doubts about whether Chinese tech firms can keep raising money through public listings.
What does the STAR50 selloff mean for U.S. investors?
The sharper question is what comes next on this side of the Pacific. Yahoo Finance frames this week’s Microsoft, Alphabet and Amazon earnings as a test of the American AI trade. Investors are also asking whether the world’s most valuable company deserves its crown.
Three things to watch next:
- Whether the state-backed buying keeps coming.
- Whether any U.S. action on Chinese data-center equipment materializes.
- Whether Chinese tech IPO plans start getting shelved.
And the question hanging over Wall Street is the one nobody wants to answer out loud. If a 100-times market can lose this much this fast, what does a 30-times market do when earnings disappoint?
