Capital expenditure on artificial intelligence (AI) in China is forecast to reach 600 billion yuan to 700 billion yuan (US$84 billion to US$98 billion) this year, according to a Bank of America (BofA) report, as the mainland and the United States race for primacy in the vital technology.
That would represent as much as a 48 per cent overall growth this year for China’s AI capex from 2024, BofA Securities’ co-head of China equity research Matty Zhao, who also heads Asia-Pacific basic materials, oil and gas research at BofA Global Research, said in an interview on Monday.
Government investment is expected to account for up to 400 billion yuan of total AI spending on the mainland, according to the BofA report published earlier this month. China’s major internet companies, meanwhile, are projected to contribute as much as 172 billion yuan.
Spending attributed by BofA to the country’s large telecommunications network operators and special-purpose bonds makes up the rest of the forecast AI capex.
The predicted surge in China’s AI spending in 2025 reflects the heightened enthusiasm for the technology on the mainland on the back of DeepSeek’s success.
The Hangzhou-based start-up generated worldwide attention earlier this year after consecutively releasing two advanced open-source AI models, V3 and R1, developed at a fraction of the cost and computing power that major tech firms typically require for large language model (LLM) projects. LLM refers to the technology underpinning generative AI services such as ChatGPT.
Following DeepSeek’s accomplishment, Chinese Big Tech companies, including Alibaba Group Holding and Tencent Holdings, announced plans for increased AI investments, BofA’s Zhao said. Alibaba owns the South China Morning Post.
E-commerce giant Alibaba in February unveiled its own 380 billion yuan capex plan for computing resources and AI infrastructure over the next three years. Tencent also announced efforts to ramp up AI investment, after the company’s fourth-quarter 2024 capex nearly quadrupled year on year to 36.6 billion yuan.
The Chinese government, meanwhile, has pushed for the buildout of new AI data centres, according to Zhao.
China is expected to step up its AI infrastructure development to match the US government’s commitment to the Stargate Project, which plans to plough up to US$500 billion over the next four years into an advanced network of data centres in the US.
While US spending on AI is geared heavily towards information technology hardware such as semiconductors, a larger portion of China’s AI capex will go to building data centres and the energy infrastructure to support those facilities, according to Zhao. She added that energy resources represent an advantage for China.
To fuel the country’s growing AI sector, Beijing earlier this month released an action plan that coordinates the development of data centres with green energy infrastructure in certain areas to meet the power requirements of these high-performance computing facilities.
That buildout in China will drive demand for a range of resources, Zhao said.
Demand for copper and power equipment will see a compound annual growth rate (CAGR) of nearly 20 per cent in volume from 2024 to 2030, while liquid cooling solutions will see a CAGR of 57 per cent in the coming year, according to the BofA report.
As China deals with US trade curbs on advanced Nvidia AI processors and waits for Huawei Technologies’ next chip breakthrough, the country continues to roll out computing facilities for AI development at a fast pace, Zhao said.
“Obviously China and the US are competing with each other,” Zhao said. “Since a lot of the chip bottleneck depends on technology development, I think what the [Chinese] government can do is what they’re good at first.”