China AI Stocks Fall as Bubble Deflates

Chinese AI Stocks Push CSI 300 Into Bear Market as $480 Billion Wipes Out China’s stock market is being pulled into a bear market not by property or consumer weakness this time, but by a violent unwind in the country’s own AI trade. The CSI 300 has fallen about 20% from its June peak and 14% this year, with the selloff concentrated in innovation-heavy benchmarks and the names tied most directly to artificial intelligence. The Shanghai STAR 50 has dropped nearly 40% from its July high, while Shenzhen’s ChiNext has lost 36%. On Thursday alone, Knowledge Atlas Technology fell 7.1%, MiniMax 6.5% and SMIC 6.2%. AI chip designers MetaX and Moore Threads have more than halved since July, while memory maker CXMT is down 27% from its debut close and 36% from its highs. The decline matters because it marks a shift in how investors are pricing China’s growth story. For years, state backing, cheap funding and policy support helped push money into sectors deemed strategic. Now the market is applying the same logic that has already crushed Chinese solar, battery and electric-vehicle margins: too much capital, too much capacity and not enough demand to sustain profitability. The question is no longer whether China can build AI infrastructure, but how many companies can survive the price war that follows. That distinction is important for the broader economy. China is already dealing with industrial overcapacity, weak domestic consumption and a property sector still under repair. A collapse in AI valuations risks repeating the same pattern seen in other favored industries, where investment surges ahead of end demand and then destroys returns across the supply chain. The pressure is especially acute because much of China’s AI buildout has been financed by public money, state-backed funds and recent IPOs in businesses that are not yet profitable. By contrast, the U.S. AI boom has been driven by mega-cap cloud companies with robust cash generation, which changes the mechanism of risk. In China, the threat is falling prices and excess capacity. In the U.S., the concern is whether the biggest technology firms can keep funding enormous data-center spending without straining their balance sheets. Meta’s latest figures underscored the scale of that capex race, and Alphabet, Microsoft and Amazon are due to report in coming days. For investors, the selloff is a warning that China’s AI trade is moving from narrative to fundamentals, and the market is unwilling to pay up for strategic ambition alone. The loss of roughly $480 billion in market value across AI-related names underlines how quickly sentiment can reverse once earnings and cash flows fail to catch up with expectations. The yuan’s roughly 9% rise this year, meanwhile, suggests capital is not fleeing China wholesale so much as being repriced away from one crowded bet. There are some signs this is not yet a full-blown macro capitulation. The currency strength argues against a broad exodus of foreign capital, and the Shanghai Composite has held up better than the most speculative segments. But that does little to change the message for stock pickers: in China, AI is no longer a one-way trade. The next catalyst will be whether policymakers or corporate earnings can slow the valuation reset before it spreads beyond the sector. | Entity | Gains | Loses | |--------|-------|-------| | Value investors | Lower entry valuations | Momentum traders | | Consumers of AI hardware | Potential price relief | Chipmakers and model builders | | Chinese policymakers | Opportunity to reprice excess | Tech sector multiples | | U.S. cloud giants | Relative valuation support | China’s listed AI names |