Recently, the Chinese Communist Party has been heavily investing in artificial intelligence (AI). Former Governor of the People’s Bank of China, Zhou Xiaochuan, mentioned the need to study and prepare for the potential financial turmoil resulting from failed AI investments.
According to mainland Chinese media reports on October 21st, Zhou Xiaochuan stated during the 2026 Bund Summit that with the massive flow of funds towards AI and related infrastructure, a situation where a portion of these investments fail leading to financial turmoil could potentially bring about a crisis. In such a scenario, massive economic support actions would be needed along with assistance to some financial institutions.
Zhou Xiaochuan emphasized, “There are several directions that can help with AI applications and enhancing productivity, but this also comes with risks.” Currently, a significant amount of investments are directed towards data center construction. However, the extent to which these AI large models and applications rely on the computing power of data centers in the future remains uncertain.
Public reports reveal that the Chinese government has planned to invest an additional 4 trillion yuan in computing network construction during the “14th Five-Year Plan” period (2026-2030). Currently, the national average GPU utilization rate in intelligent computing centers (AI computing centers) is less than 30%, and the utilization rate of enterprise-level general-purpose computing centers is only 10% to 15%. Industry analysts point out that an intelligent computing center with an annual operating cost of 30 million yuan and a utilization rate of less than 30% poses a potential credit risk.
The current wave of AI investments in China is being driven by the Chinese state-owned capital system. Recently, the establishment of the Shaanxi Social Security Technology Innovation Equity Investment Fund (Phase I) of tens of billions of yuan was officially announced, focusing on investments in hard technology and the AI industry. This fund includes investments from China Bank Asset Management, Chang’an Huitong, Xi’an High-tech Golden Control, and other local state-owned enterprises.
The 2026 Shanghai Pudong New Area “Venture Capital Industry White Paper” indicates that the artificial intelligence industry is leading in financing. Simultaneously, entities with state-owned backgrounds like the Shanghai Jiaotong University AI Future Fund, Wuxi Data Group, jointly invested hundreds of millions of yuan in local smart enterprise “Quadra Intelligence”.
The surge of AI investments in China has attracted widespread attention, with institutions and experts highlighting underlying risks such as the low utilization rate of data center computing power, the incompetence of local government leaders in managing AI investments, exacerbation of economic disparities, and water resource scarcity issues.
A report by the British Broadcasting Corporation (BBC) previously noted that from a corporate perspective, government funding has artificially supported the market, yet it has also influenced market dynamics. Local governments have become “shareholders,” gaining not only equity but also influence over corporate decision-making. Entrepreneurs are faced with the challenge of balancing between sustaining on government funding and preserving autonomy in decision-making.
In addition, the report suggests that it isn’t easy for local Chinese governments to rely on AI investments to boost fiscal revenue. Due to the rapid contraction of land finance, equity-based finance might not provide immediate solutions. While land sales generate immediate revenue, equities usually have a return cycle of 7 to 10 years, requiring initial investments without immediate returns; profits often depend on the success of initial public offerings (IPOs) or mergers and acquisitions.
In a research report, Citibank pointed out that the AI-driven manufacturing boom continues to expand, but domestic consumer demand is weakening, leading to a growing disparity. This divergence in economic signals within China poses contradictory challenges.
On October 13th, James Gorrie, a columnist for English Epoch Times, raised another concern: the substantial water requirements of artificial intelligence. As data centers are being rapidly built across the country, 72% of these centers are located in severely water-scarce regions. This raises a fundamental question – can the Chinese government expand its AI infrastructure without worsening the existing water resource crisis?
The article suggests that water resources may become one of the most critical constraints in China’s AI expansion in the 2030s. China lacks the capacity to mass produce freshwater to meet the fundamental needs of AI, industry, agriculture, and public water usage.
Furthermore, Rystad Energy predicts that between 2025 and 2030, the power consumption of Chinese data centers will increase at an annual rate of about 19%.