Zhou Xiaochuan, former governor of the People’s Bank of China, said on October 20 that while several major directions could help AI applications boost productivity, they also entail risk. Speaking at a roundtable at the 2026 Bund Summit on AI and the productivity revolution, he noted that heavy investment is flowing into data center construction, but it is uncertain how far future large AI models and applications will depend on data center computing power.
Zhou divided AI-related investment into three categories: large models along with the computing power and data centers used to train them; AI applications, which China is paying closer attention to; and supporting industries such as semiconductors, electricity and liquid cooling. He considered the third category safe, since AI development will require water and electricity regardless of how it unfolds.
He said productivity gains would come mainly from the application layer, citing lightweight models that shrink large models for use in cars, laptops and smartphones, and distillation. Although some worry distillation infringes intellectual property, Zhou argued it represents an upgrade in intelligence, building on existing knowledge, and could bring investors generous returns. He also named model iteration, AI-related semiconductors and large simulation models as other developments.
On international cooperation, Zhou said rapid change makes governance frameworks difficult, but at least three steps are feasible: restricting military uses, combating hackers and fraudsters, and protecting personal privacy. Effects on employment, income distribution and social structure remain at the observation stage, with no consensus yet.
On finance, Zhou urged letting price mechanisms respond to shifting global supply and demand rather than relying excessively on administrative measures. He also warned that if some of the enormous sums flowing into AI and related infrastructure fail, financial turbulence or crisis could follow, requiring large-scale economic support and rescues of some financial institutions. He called for research and contingency planning, with suitable tools stored in the policy toolbox.
Source: Yicai.com, October 21, 2026
https://www.yicai.com/news/103401935.html