Zhou Xiaochuan, former governor of the People's Bank of China, said on October 20 at the 2026 Bund Annual Conference that China should prepare contingency plans for financial risks arising from failed AI-related investments. On October 22, Chinanews reported his further view that China's existing financial tools remain effective against AI risks.
1. Three categories of AI investment, three risk profiles
Zhou divided financial-sector AI investment into three categories: large models plus training compute and data centers; AI applications — which he called the main source of productivity gains and China's key focus; and supporting industries such as semiconductors, power and liquid cooling. "Investing in these is safe, because no matter how AI develops, water and electricity are always needed."
2. The probability of failure under huge capital flows
Some investments will inevitably fail. But with such enormous capital flowing into AI and its infrastructure, a significant wave of failures could trigger financial turbulence and even a crisis, requiring large-scale action to support the economy and bail out some institutions. "Therefore, we should conduct research and prepare plans, stocking the policy toolbox with corresponding instruments," Zhou said.
3. When governance frameworks lag, price mechanisms step in
Zhou noted AI changes too fast for inter-governmental governance frameworks to keep up, but at least three things can be done: controlling military uses, cracking down on hackers and fraudsters, and protecting personal privacy. On structural change, he stressed letting price mechanisms — rather than administrative measures — reflect and adapt to the shifting industrial structure.
Key points from Zhou's Bund appearance
· Venue: "AI and the Productivity Revolution" roundtable, Oct 20
· Core warning: failed AI investments could spark financial turbulence; prepare policy tools in advance
· Three actionable items: military-use control, anti-fraud, privacy protection
From the debate over Nvidia's ROI to Microsoft pulling out of a UK data-center project over power delays, markets are repricing the sustainability of AI capex. Zhou's reminder points to a plain fact: the dividend of innovation and the cost of bubbles are carried by the same financial institutions, so risk planning must keep pace with the investment boom — the same principle behind NineZenith Tiandun's "model risk and business risk in one frame" approach for financial clients.
(Information synthesized from The Paper, Chinanews, Yicai and other public reports)