Sunita Somvanshi
China’s AI stock bubble burst has wiped $480 billion from mainland equity markets in under four months, even as Wall Street priced technology stocks near record highs. Mainland China’s STAR50 index — the benchmark for domestic AI and semiconductor leaders — has dropped 40% from its peak, pulling the broader CSI 300 more than 20% below its June highs and into bear market territory.
At its height, the STAR50 traded at roughly 250 times trailing earnings, a level requiring decades of compounding growth to justify. That multiple has since compressed to approximately 100 times earnings — still extreme by global standards, but down sharply. The CSI 300’s decline reflects losses spreading beyond pure-play AI names to semiconductor equipment suppliers and cloud providers. Regulators contributed by cracking down on leveraged positions and high-frequency trading strategies that amplified the rally on the way up. Goldman Sachs research shows AI-related stocks now represent 42% of China’s combined domestic and Hong Kong market capitalization and account for 57% of A-share and H-share trading turnover by volume — a concentration that makes the sector’s volatility felt across nearly every mainland portfolio.
For investors holding emerging-market ETFs or global semiconductor positions, the $480 billion selloff changes both valuation benchmarks and IPO expectations. Even after falling 40%, the STAR50 still trades at 100 times trailing earnings — more than triple the Nasdaq 100’s 30x multiple — leaving material downside risk in broad index funds tracking Chinese technology. Institutional buyers are moving toward selective supply-chain names such as Innolight and CXMT or rotating into state-backed value equities rather than buying the broad index.
Retail Wipeout vs. State ETF Intervention
Wang Zhuo, a partner at Shanghai Zhuozhu Investment Management, has identified the domestic chip manufacturing price war as a structural driver beneath the crash. As Chinese fabless designers scaled up following US export restrictions on advanced semiconductors, component prices across the supply chain collapsed well before the broader market discounted the margin damage. The STAR50’s original rally had priced hardware suppliers for top-line revenue growth without accounting for the competitive pressure that state-backed capacity expansion would impose on profits.
The IPO pipeline has frozen in parallel. Hong Hao, chief investment officer at Lotus Asset Management, has pointed to the post-listing performance of robotics manufacturer Unitree as a warning. When a closely watched domestic AI debut underperforms, appetite for new AI paper dries up across the sector, and the queue of mainland companies seeking regulatory listing approval stalls as institutional allocation committees reassess valuations.
China’s “National Team” — state-run funds authorized to purchase domestic equity ETFs during periods of market stress — provided partial insulation to the broader index. Bloomberg data shows National Team ETF inflows reached record levels at the peak of the STAR50 selloff, absorbing retail selling pressure and forming informal price floors within the CSI 300. The intervention stabilized the broader market without reversing the AI sector’s decline; state buying was directed toward diversified blue-chip positions rather than the high-multiple AI names central to the crash. William Bratton of BNP Paribas Securities has described the current intervention pattern as more selective than in prior cycles, a signal that officials are comfortable letting AI-specific valuations compress further without direct sector support.
A 40% decline typically leaves a sector cheap. The STAR50 is an exception. At roughly 100 times trailing earnings — down from a peak of 250x — it still trades at more than triple the Nasdaq 100’s 30x multiple. With AI names generating 57% of all A-share and Hong Kong turnover, over half the market’s liquidity was crowded into a sector still priced for perfection as domestic chipmakers entered a margin-compressing price war.
Why did China’s AI stock bubble burst in 2026? China’s AI rally collapsed under extreme valuations — the STAR50 index reached 250 times trailing earnings — combined with US semiconductor export curbs, domestic price wars among chip suppliers, and regulator crackdowns on leveraged and high-frequency trading. The resulting four-month selloff erased $480 billion from the STAR50 and pushed the broader CSI 300 into a bear market.
The split between domestic retail exits and foreign institutional positioning has rarely been this pronounced. HSBC research shows Asia institutional funds remain net overweight on Chinese equities. Goldman Sachs retained Innolight and CXMT on its October conviction buy list. The $480 billion crash has separated indiscriminate sellers from institutions selecting specific positions in the surviving supply chain.
For related context, see Karmactive’s earlier coverage and latest explainer.
