BEIJING: Has China’s artificial intelligence bubble finally burst? A once-soaring rally in Chinese AI stocks has lost much of its momentum, with a sharp sell-off raising questions about whether enthusiasm for the technology has pushed share prices beyond what companies' earnings could justify.

But while the decline has fuelled debate over whether an AI investment bubble is deflating, analysts told CNA that falling share prices do not necessarily signal a setback for China’s broader AI ambitions - or suggest that other markets are headed for a similar correction.

As of Thursday (Oct 29), China’s STAR 50 Index, which tracks 50 major tech and innovation companies on the Shanghai Stock Exchange, had fallen more than 40 per cent over four months, wiping out some US$480 billion in market value.

The scale of the decline has brought renewed scrutiny as to whether heavy spending on AI can translate into higher profits, and what the reversal means for China’s tech ambitions and the wider global AI investment boom.

“The sell-off appears to be primarily a repricing of expectations, but it is also prompting more scrutiny of monetisation, profitability and capital efficiency,” said Stephen Wong, an executive committee member of the Greater China Chapter of the Chartered Alternative Investment Analyst (CAIA) Association, a US-based global professional body for the alternative investment industry.

While falling share prices could make it harder for some Chinese firms to raise funds, the downturn is unlikely to derail Beijing's broader AI ambitions, analysts said - which continue to benefit from government support.

WHY CHINA'S AI STOCK RALLY IS LOSING STEAM The sell-off follows a roughly two-year rally in Chinese equities, fuelled in part by enthusiasm for AI that sent technology shares soaring, particularly in sectors such as semiconductors and robotics. Despite the steep correction, valuations remain elevated. An illustration of a robot touching forex and stock market charts displayed.

As of Oct 26, the STAR 50 was still trading at around 100 times the combined earnings of its constituent companies over the past 12 months - down from a peak of nearly 250 times, according to Reuters.

That suggests that investors are still paying a substantial premium for future growth - leaving shares vulnerable if earnings fail to meet expectations.

Kelvin Wong, senior market analyst at online trading platform OANDA, said Chinese equities have been particularly exposed to a reversal in investor sentiment.

Citing Goldman Sachs figures, Wong noted that AI-related stocks accounted for about 42 per cent of market capitalisation and 57 per cent of trading activity across mainland-listed A shares and Hong Kong-listed H shares.

“That level of concentration leaves the market particularly vulnerable when investors begin taking profits,” he said.

Beyond speculative trading, analysts also pointed to a more fundamental concern: whether China’s AI boom can translate into sustainable profits.

“It goes beyond the correction of excessive valuation,” said Gary Ng, senior economist for Asia Pacific at Natixis.

“Investors are questioning whether demand can turn into higher profits, especially as capex continues to climb,” he said, referring to capital expenditure, or spending on equipment and infrastructure.

OANDA’s Wong pointed to intense competition among Chinese tech companies and firms - warning that aggressive pricing could squeeze profit margins even as demand for AI computing capacity grows.

Geopolitical uncertainty has also added to the pressure, analysts said - with Chinese suppliers facing the prospect of US restrictions on their equipment being used in American data centres, which could potentially limit access to overseas customers.

Christopher Beddor, deputy China research director at research firm Gavekal Dragonomics, said that the downturn was part of a wider regional sell-off and a recent drain of market liquidity.

“There are also real concerns about the earnings prospects for some of China's AI companies as they struggle to monetise open-weight models,” he said, referring to AI models whose underlying parameters are publicly available for others to use and adapt.

THE REAL TEST FOR CHINA’S AI AMBITIONS Despite the sharp decline, analysts cautioned against mistaking a stock market downturn for a broader setback to China’s AI ambitions.

The more immediate risk, they said, is whether weaker investor confidence could make it harder for Chinese technology firms to raise capital - particularly those with high spending and limited revenues.

“It will make fundraising harder in terms of valuation and delay, especially for firms that are burning cash without immediate revenue,” said Ng from Natixis.

But he added that strategically important firms would likely continue to benefit from substantial state support, leaving smaller players more vulnerable to the downturn.

Wong from the CAIA Association said the correction could also make investors more selective, potentially slowing some projects and encouraging consolidation among companies with weaker business models.

China's AI industry could still draw on government-backed funds, bank lending, corporate bonds and private investment, Wong said - reducing its dependence on stock market fundraising.

“The key distinction is between supporting the development of an industry and generating an attractive return for every investor,” he added.

People walk past a billboard promoting the 2024 World Artificial Intelligence Conference (WAIC) and High-Level Meeting on Global AI Governance in Shanghai, China, on Jul 3, 2024. (File photo: AP/Andy Wong)“Government support can facilitate infrastructure, research and capacity building, but it cannot by itself ensure that every company develops a sustainable business model or that every security represents good value.”The impact on financing could become apparent over the next quarter or two, said Beddor from Gavekal Dragonomics - who expects the downturn to dampen stock market listings.

But for China’s longer-term AI ambitions, Beddor pointed to a bigger hurdle: access to advanced chips.

“The real constraint on China's AI ambitions is not financing but access to chips,” he said.

US export controls have restricted Chinese firms' access to some of the most sophisticated foreign-made AI processors, which are crucial for training and running increasingly powerful AI models.

While Beijing has stepped up efforts to develop homegrown alternatives, Chinese chipmakers still face challenges in producing advanced AI processors in sufficient quantities to meet demand. COULD THE AI SELL-OFF SPREAD?

China's AI stock market rout has raised a question for investors elsewhere: Could a similar reckoning be looming, particularly in the US, where technology giants have poured billions into artificial intelligence?

OANDA’s Wong described the Chinese sell-off as an early warning of valuation and financing risks, rather than a sign that global demand for AI was faltering.

“China's experience demonstrates how quickly sentiment can reverse when investors begin questioning the earnings assumptions underpinning an extended rally,” he said. But analysts cautioned against drawing direct parallels between the two markets.

Major US technology companies like Microsoft, Alphabet, Amazon and Meta have established businesses generating substantial revenue and cash flow, Wong said - giving them a stronger financial base to support AI investment. Beddor was also more cautious about drawing parallels between the two markets.

“China's stock market marches to the beat of its own drum, which is quite different from US markets and sometimes even the underlying Chinese economy,” he said.

“But it's hard to do a clean read-across from China's stock market to the future of US markets.”Elsewhere in Asia, the contrasting fortunes of AI-related companies underscore why the sector cannot be treated as a single investment category.

In South Korea, chipmakers SK Hynix and Samsung Electronics reported record quarterly operating profits this week, buoyed by strong demand for memory chips used in AI systems.

People queue to visit the SK Hynix booth at the 28th Semiconductor Exhibition in Seoul, South Korea, on Oct 16, 2026.

For investors, the focus is increasingly on whether rising AI spending is matched by growth in revenue and profits.

Wong from OANDA pointed to disappointing earnings forecasts from major US technology companies, tighter financing conditions and weaker semiconductor demand as potential triggers for a broader reassessment of AI stocks.

Wong from the CAIA Association said investors should also look beyond listed technology shares, as expectations about AI demand are reflected in investments ranging from data centres and corporate bonds to private credit and power infrastructure.

“These exposures may react at different speeds, so apparent diversification by asset class may still conceal concentration in the same economic driver,” he said.

He also stressed that investors needed to distinguish between the long-term prospects of AI and whether individual companies were worth their current valuations.

“It is possible to be right about the long-term theme but wrong about the security, the capital structure or the valuation.”