China’s AI Stock Bubble Just Lost $2.4 Trillion

China’s bet on AI is costing its stock markets a lot of money. According to Hacker News, which points to a report built on Bloomberg data, about $2.4 trillion has been wiped off Chinese stocks since May. The Shanghai STAR 50 Index, the country’s main tech and innovation benchmark, has fallen 40% since June.

That’s not a small correction in one corner of the market. Goldman Sachs estimates that AI stocks make up more than 40% of total market value in China and Hong Kong. When AI shares drop, the whole market drops with them. Other major indexes are down about 20% over the same period.

Why Investors Are Pulling Back

The sell-off isn’t about whether Chinese AI works. It’s about whether anyone in China will pay for it. Dan Wang, China director at the political risk consultancy Eurasia Group, sums up the mood: “This year, the profit margin is not as good as expected. And the US is raising rates. And China’s domestic demand is so weak.”

Several pressures are piling up at once:

  • Thin profits. Chinese consumers are still reluctant to spend, so many tech companies can’t turn growth into earnings.
  • US restrictions. Washington has banned Chinese humanoid robots on national security grounds. It’s reportedly weighing a ban on Chinese components in data centers too.
  • Dependence on exports. Chinese firms can’t sell enough at home to make up for losing foreign markets.
  • Higher US interest rates. They pull global capital away from riskier emerging-market tech.

Wang describes a “general anxiety” among tech companies about whether the state-backed model can work when consumer demand is this weak.

The Robotics Problem

Beijing has gone all in on embodied AI, meaning robots that put AI to work in the physical world. The pitch is strong on paper. Robots could help with China’s aging population and shrinking workforce.

The market tells a different story. Most robotics demand comes from government agencies and research institutions, not households. Humanoids and robot dogs look great at trade shows and in entertainment. At home, though, they’re still too expensive and not very useful.

Shares in Unitree, China’s leading robot maker, have halved since the company listed in August. “It doesn’t look like embodied intelligence could turn into profits in the mass market,” Wang said.

What stands out here is the gap between policy and demand. The state can fund supply. A new 60 billion yuan state-backed AI fund started investing in domestic developers this month. But the state can’t make people buy robot dogs.

A Tale of Two Capital Markets

The contrast with the US is hard to miss. The same week Chinese tech stocks kept sliding, Anthropic filed paperwork for an IPO that could raise as much as $100 billion. That would make it one of the biggest stock market debuts ever. China has nothing like that coming.

Beijing’s attempts to keep money at home haven’t helped much either. New limits on overseas investment take effect December 1, but they haven’t brought fresh money into domestic stocks.

Don’t expect a rescue soon. Party leaders are meeting for the fifth plenum, which sets up next year’s 21st Party Congress, where Xi Jinping is expected to get a fourth term. Before then, nobody wants to sign off on risky stimulus. “What if it doesn’t work out? Then whoever came up with that policy, that guy would be held accountable,” Wang said.

Not Every Sector Is Hurting

Wang still sees some bright spots. Investors remain fairly upbeat about AI research and satellites. Satellites have military ties, which suggests long-term state support. Globally, she expects the AI boom to keep going for at least another year.

What This Means for AI Businesses

This is significant because it shows that heavy state funding can’t stand in for real demand. Here’s what practitioners and businesses should take from it:

  1. Follow the revenue, not the capex. Big infrastructure spending only pays off when customers pay. That applies in the US as much as in China.
  2. Plan for decoupling. If the data center component ban goes ahead, supply chains will split further. Check your hardware and vendor exposure now.
  3. Be careful with embodied AI hype. Consumer robotics is still years away from mass-market economics. Industrial and government uses are where the money is today.
  4. Watch the IPO window. Huge US listings could suck up global investor attention and make capital even scarcer for non-US AI firms.

In my view, the bigger story is the gap between two AI economies. One has paying customers and deep capital markets. The other has state money and cautious consumers. Until Chinese households start spending, Beijing’s AI push will keep running into the same wall. More details are available at the original source.

Scroll to Top