China profits jump 25.7% but AI costs sink stocks

The numbers tell two stories at once. Onshore-listed Chinese companies just posted a 25.7% profit surge in the three months to June, the fastest clip in nearly five years. That headline should have sparked a rally. Instead, the CSI 300 index has fallen about 9% this quarter, and the tech-heavy Star 50 has cratered 29%. The culprit? Investors have flipped their view on artificial intelligence spending, now seeing it as a cost burden rather than a future payoff.

This isn't a typical earnings miss. The profit growth is real, broad-based, and concentrated in sectors tied to AI infrastructure, chip manufacturing, and data centers. But the market's reaction reveals a deeper anxiety: companies are pouring billions into AI capacity with no clear timeline for returns. When the spending was a promise, investors rewarded it. Now that it's a line item on income statements, they're punishing it. The Star 50, which hosts many of China's most AI-exposed firms, has borne the brunt of this shift.

The divergence between earnings and stock prices is a classic sign of a regime change. Historically, strong profit growth lifts indices. But when the growth itself is driven by capital expenditure that may not generate near-term cash flow, the market starts to question sustainability. In China's case, the government has pushed hard for AI self-reliance, leading to a surge in domestic chip orders and server builds. That's boosted profits for suppliers, but it's also created a glut of capacity that could depress margins later.

Consider the mechanics. A 25.7% profit jump means companies are making more money, but if that money is being reinvested into AI projects with uncertain payback periods, the market sees risk. The CSI 300's 9% decline suggests a broad de-rating, not just in tech. Even traditional sectors are being dragged down by the fear that AI spending will crowd out other investments or lead to a bubble burst. The Star 50's 29% drop is more severe because it's concentrated in the very companies making the biggest AI bets.

This is not unique to China. Globally, investors have started questioning AI capex from US hyperscalers too. But China's situation is amplified by geopolitical constraints. Sanctions on advanced chips force domestic firms to use less efficient alternatives, raising costs and lowering returns on AI investments. The profit surge, then, is partly a function of government subsidies and forced localization, not pure market demand. When the subsidies taper or the technology gap widens, those profits could evaporate.

For executives and investors, the takeaway is that earnings alone no longer move markets. The quality of earnings matters, especially the split between operational efficiency and capital-intensive expansion. A company that doubles profits by cutting costs is viewed differently than one that doubles profits by spending heavily on AI. The latter is now treated with suspicion, as the market demands evidence of revenue generation, not just cost absorption.

The second half of the year will be telling. If AI spending starts translating into tangible products or services with paying customers, the sell-off could reverse. But if the capex cycle continues without clear monetization, the Star 50's 29% drop might be just the beginning. The CSI 300's 9% decline could deepen, dragging the entire market down despite strong headline earnings.

For now, the message is clear: in China's equity markets, AI has shifted from a growth narrative to a cost narrative. The 25.7% profit jump is real, but it's being overshadowed by the bill for tomorrow's technology. Investors are voting with their feet, and they're saying that promises don't pay dividends. The companies that can show AI investments generating actual returns will be rewarded; those that can't will face continued selling pressure. This is a moment for disciplined capital allocation, not blind faith in the AI story.