Chinese financial regulator warns AI investment is too concentrated

A senior official at China's financial regulator says global AI investment is "overly concentrated" and is pushing up borrowing costs for traditional industries. Xiao Yuanqi, vice head of the National Financial Regulatory Administration, made the remarks on Tuesday at the Hong Kong Monetary Authority's Global Financial Leaders' Summit, according to Reuters.

To be clear, this is a warning from the podium. No deal, rule or new funding programme came with it.

The argument Xiao made

Reading the Reuters report, his case runs in a straight line. A flood of capital into the tech sector and its leading firms has raised leverage there, which could fuel expansion that's too rapid. That concentration then crowds out traditional industries: it lifts their borrowing costs, weakens their ability to reinvest and expand production, and raises credit risk.

His ask was for policymakers to prepare frameworks aimed at this AI-driven shift, covering both macroeconomic management and the prevention of financial risks. Reuters didn't report any specific measures.

One small discrepancy on titles: Reuters calls him vice head, while the HKMA's 2026 summit programme lists him as Vice Minister of the National Financial Regulatory Administration. It's the same person and the same body.

Who pays and who gains

Going by Xiao's own logic, the split looks like this:

• Who pays: traditional-industry borrowers, which face higher borrowing costs and a heavier financial burden.
• Who gains, with a catch: tech firms and the leading AI players that are drawing in capital. Xiao said that same inflow has raised their leverage.
• Who carries the risk: lenders, since he linked the concentration to higher credit risk.

I think it's best read as one regulator's framing rather than a measured result. Reuters didn't report any data from the speech to back the borrowing-cost claim.

The same pattern shows up elsewhere

Other reporting points the same way. A Bloomberg story in the Taipei Times says surging AI borrowing demand has Taiwanese banks rationing credit, raising loan rates and squeezing lending to non-tech firms. I think that's a close match for what Xiao described.

In the US, a Yahoo Finance piece argues AI appears less sensitive to rising interest rates, so the strain lands on other borrowers. It also cites the Financial Times, using Morgan Stanley data, saying global AI-debt issuance in September was $23 billion. That's below June's record of $113 billion, so the borrowing binge may be slowing.

Beijing has voiced similar worries at home. In an official explanation published on October 9, the NDRC warned against repetitive industrial construction and poorly planned investment, according to Tech Wire Asia. Xiao's comments probably fit that line, though they're about global capital flows, not China's alone.

Next steps to watch

Nothing concrete is scheduled. Xiao called for policy frameworks, but no regulator has said what one would contain. I'd guess any follow-up shows up first in lending guidance or credit-risk monitoring rather than anything aimed at AI firms directly, but that's speculation.

For anyone tracking the AI market, the useful signal is probably who's saying it. A financial regulator is worried about credit, not model quality. Honestly, I'd pay more attention to the borrowing-cost data than to the speech itself.

One limit on the sourcing: I could only find Xiao's remarks in Reuters reporting and copies of it, not a full transcript. The quotes here are what Reuters chose to publish.