China’s AI stocks lose $480 billion as investors reassess the boom

A sharp repricing

Reuters reports that the STAR50, which includes Chinese AI and semiconductor companies, has fallen 40% since the start of the four-month slide. Fund manager Dong Baozhen told Reuters the stock market had become a speculative boom and that the bubble was now bursting.

The losses extend beyond the index:

Money is still moving

The selloff has not stopped large investments. Bloomberg reports that DeepSeek recently raised $15 billion, one of the largest funding rounds in China’s history. Huawei has also received a $2.3 billion loan from Chinese banks to buy new AI chips.

But China’s investment environment has constraints that do not map neatly onto the usual bubble narrative. The United States is trying to limit China’s role in global competition through tariffs and consideration of a full ban on Chinese companies supplying equipment to data centers. Zhongji Innolight, whose optical transceivers carry the vast volumes of data used by AI at the speed of light, is among the companies affected by that pressure.

That leaves two possible sources of strain: investors may have grown too optimistic, and Chinese firms may be finding it harder to sell AI products overseas. The second deserves more attention than it gets in a story framed only as a market correction.

The return question

Goldman Sachs estimates AI stocks account for about 42% of China’s market capitalization, a concentration comparable to the scale of investment in AI and related industries in Western economies. There, too, investors are confronting a basic problem: a clear return on AI spending remains hard to see.

Anthropic disclosed losses of $80 billion in its IPO documents and plans to spend another $1 trillion on data centers over the next decade. Those figures could deepen concerns that even the largest players are far from profitability.

Reuters also reports that state-linked investors sold shares during market rallies, part of a government campaign to reduce volatility. I think that may have helped restrain prices, but it does not answer the harder question: whether demand for AI products can support the spending. If overseas sales are constrained as well, lower valuations may be reflecting more than a bubble deflating.