The Bursting of Beijing’s AI Bubble: How $2.4 Trillion Evaporated from China’s Markets

Main Facts

China’s mainland stock markets have suffered an astonishing capital flight, shedding approximately $2.4 trillion in market value since May. To put this staggering sum into perspective, the lost wealth is greater than the entire annual economic output of South Korea. This massive economic contraction has been catalyzed by a sudden, severe loss of investor confidence in Beijing’s heavily promoted artificial intelligence boom—specifically, mounting fears that massive capital expenditures in AI infrastructure and model development will never translate into sustainable, long-term corporate profits.

The crisis is systemic because of AI’s outsized footprint in Chinese equity. According to recent data from Goldman Sachs, artificial intelligence-related companies now account for an overwhelming 42% of the total market capitalization of Chinese shares listed across mainland exchanges and Hong Kong, while driving a staggering 57% of daily trading volume. Consequently, when the AI sector experiences a downturn, the entire Chinese equity apparatus is dragged down with it.

The repercussions have been swift and severe. The CSI 300 index, which tracks the 300 largest listed companies on mainland exchanges, officially entered a bear market on October 22, plunging more than 20% from its previous peak. Meanwhile, the Shanghai STAR 50 index—a high-flying benchmark composed of China’s most prominent tech, chipmaking, and robotics enterprises—nosedived by 40% in less than four months, obliterating roughly $480 billion in value.

Despite these punishing corrections, valuations remain remarkably high compared to Western standards. The STAR 50 still trades at roughly 100 times earnings, meaning investors are paying for a century’s worth of current profits at present rates—though this is a steep descent from the dizzying valuation of nearly 250 times earnings seen at the sector’s peak. For comparison, the tech-heavy U.S. benchmark Nasdaq 100 hovers around a much more modest price-to-earnings ratio of 30.

Chronology of a Collapse

The unraveling of China’s AI equity dream did not happen in a vacuum; it is the culmination of a rapid, speculative boom-and-bust cycle driven by technological breakthroughs, sweeping geopolitical interventions, and shifting market psychology.

• January 2025: The domestic and global AI landscape was radically disrupted when Chinese startup DeepSeek released an exceptionally cheap yet remarkably powerful open-source chatbot. While framed as a triumph for Chinese engineering, the model’s low cost sent shockwaves through global markets, triggering a staggering $600 billion single-day wipeout in market value for American semiconductor giant Nvidia, whose high-end chips power the vast majority of Western AI workflows.
• July 2025: Market anxiety was reignited when another domestic AI lab, Moonshot AI, rolled out its advanced Kimi K3 model. The release triggered a fresh wave of panic selling across chip and tech stocks, proving that the hyper-competitive domestic environment was eating away at pricing power and profit margins. In the same month, external pressures intensified as the U.S. Federal Communications Commission (FCC) enacted a sweeping ban on the import of new foreign-made humanoid robots, citing unspecified national security concerns.
• August 2025: Investor appetite for high-spec tech IPOs reached a fever pitch before abruptly hitting a wall. Prominent Chinese robotics firm Unitree went public, drawing retail orders for a staggering 8,000 times the shares on offer. However, the speculative frenzy quickly soured; Unitree’s stock plummeted, losing nearly half its launch value within weeks—even after staging a modest recovery later in the year.
• September 2025: State intervention reshaped the hardware landscape. According to reports from the Financial Times, Beijing’s internet regulators directed major domestic tech giants, including ByteDance and Alibaba, to halt their procurement of Nvidia hardware entirely, forcing a hard pivot toward domestic alternatives.
• October 22, 2025: The mounting pressures culminated in a technical rout as the CSI 300 index officially entered bear market territory, marking a definitive end to the speculative high-water mark that had gripped mainland markets for much of the year.

Supporting Data and Market Metrics

The scale of the downturn is illuminated by granular financial metrics and supply chain data that reveal the depth of the speculative mania and its subsequent correction:

• $2.4 Trillion: The total equity value lost across mainland Chinese stock markets since May, eclipsing South Korea’s annual Gross Domestic Product (GDP).
• 42% and 57%: The proportion of Chinese share market capitalization and trading volume, respectively, tied directly to the artificial intelligence sector, highlighting the extreme concentration risk for investors.
• 40% Drop in 4 Months: The speed at which Shanghai’s STAR 50 index fell, wiping out $480 billion in localized tech wealth.
• 100x vs. 30x: The current price-to-earnings (P/E) ratio of China’s STAR 50 index compared to America’s Nasdaq 100. While the Chinese tech index has retreated from an astronomical peak of 250 times earnings, it remains profoundly expensive relative to Western tech counterparts.
• 60% Retracement: The steep decline experienced by crucial Chinese optical component manufacturers like Zhongji Innolight and Eoptolink from their all-time highs, heavily driven by looming U.S. restrictions on Chinese-made optical infrastructure in domestic data centers.

Official Responses and Industry Perspectives

The collapse has forced a stark reassessment among financial analysts, fund managers, and state regulators regarding the viability of state-backed technological crusades in the face of cooling consumer demand.

Dong Baozhen, a prominent Beijing-based fund manager, offered a candid autopsy of the market dynamics during an interview with financial media. “The stock market later evolved into a speculative crazy bull,” Baozhen noted, reflecting on the manic influx of retail capital into unproven tech ventures. “And now we see the bubble bursting.”

Underpinning the financial panic is a fundamental macroeconomic weakness: anemic consumer spending within China. While Beijing has continuously poured vast state subsidies and policy directives into foundational technologies—urging firms to embrace homegrown chips and domestic large language models—ordinary consumers have tightened their purse strings. This disconnect has left AI companies sitting on enormous capital expenditures without a clear, scalable path to monetization.

Furthermore, laboratories have rushed to validate the national self-reliance narrative. Z.ai, a prominent Beijing-based artificial intelligence lab, recently made waves by announcing that its latest advanced model, GLM-5.2—which rivals Western systems like Anthropic’s Claude Opus—was trained entirely on domestic Huawei Ascend chips, bypassing Western embargoes entirely. Yet, while technological self-sufficiency has been achieved on paper, market participants are increasingly asking whether these costly domestic hardware ecosystems can generate enough commercial revenue to justify their astronomical valuations.

Global Implications

The fallout from China’s AI stock market correction extends far beyond domestic borders, carrying profound implications for the global technology supply chain, geopolitical trade relations, and international capital flows.

First, the synchronized decoupling of the technological ecosystem is accelerating. As Beijing successfully forces its domestic champions—such as ByteDance and Alibaba—to completely abandon Nvidia hardware in favor of indigenous chips from firms like Huawei and domestic foundries, a bifurcated global AI economy is solidifying. Companies operating in this fractured landscape must now build models tailored to distinct, often incompatible hardware stacks, driving up research and development costs globally.

Second, Western regulatory pressure continues to close off vital international markets for Chinese technology firms. Beyond the FCC’s ban on humanoid robotics, Washington is actively advancing legislation to prohibit Chinese-made optical parts and networking components from being utilized in U.S. data centers. This regulatory pincer movement has severely damaged high-flying Chinese hardware suppliers like Zhongji Innolight and Eoptolink, whose shares remain down roughly 60% from their peaks as access to lucrative Western markets dries up.

Ultimately, the bursting of China’s AI bubble serves as a cautionary tale for global markets. It underscores the reality that even unprecedented state backing and immense speculative enthusiasm cannot permanently decouple technological ambition from the foundational laws of economics: eventually, billions of dollars in infrastructure spending must yield tangible profits, or the market will exact a brutal reckoning.