Bank of America Securities has significantly raised its forecast for capital expenditure on Chinese AI data centers (AIDC) to $327 billion by 2030 in a new report on AI infrastructure. The report systematically outlines structural investment opportunities for traditional industries like copper, PCB materials, optical fiber, and transformers entering the AI value chain, with related stocks receiving a series of buy ratings.
It has been reported that China's AIDC capital expenditure over the next five years could reach approximately 2 trillion yuan. According to market information, Bank of America expects China's AI capital expenditure to grow from around $140 billion in 2026 to $327 billion in 2030, representing a compound annual growth rate of 24%. This would account for about 20% of global AI capital expenditure by that time. Concurrently, the bank's global team has also revised its forecast for worldwide AI capital expenditure upward to over $1.7 trillion by 2030, a substantial increase from $260 billion in 2025.
Core Investment Themes
The report is structured around two primary investment themes. The first is AI materials, encompassing five categories: copper, PCB materials (copper foil and glass fiber), optical fiber, and magnetic materials/tungsten/uranium. Several high-end sub-categories within these areas are facing structural supply shortages, providing strong upward momentum for prices. The second theme is AI power supply, which includes five key opportunities: transformers, gas turbines, diesel engines, energy storage systems, and power supply units. China, with its advantages in electricity costs, grid conditions, and a robust equipment supply chain, is well-positioned to benefit deeply from the global AIDC construction wave and increase its export share.
The core logic behind this outlook is that the AI computing race is increasingly becoming a competition in power infrastructure. Global data center installed capacity is projected to expand from the current approximately 100GW to nearly 300GW by 2030. Power density per rack is expected to surge from the 10-15 kilowatts of traditional servers, following the Nvidia platform roadmap, to 100-120 kilowatts currently, and potentially exceed 1 megawatt in next-generation systems. This will drive upstream material and power equipment demand into a structural upward trend.
Power Demand Projections
According to data from the International Energy Agency (IEA), global data center electricity consumption will approach 500 terawatt-hours in 2025, representing about 1.6% of global electricity use. Bank of America estimates this figure will expand at a 22% compound annual growth rate, reaching 1,208 TWh by 2030 (approximately 3.7% of global electricity consumption).
In China, data center installed capacity is forecast to expand from 29GW in 2025 to 77GW in 2030, with corresponding electricity consumption rising from 121 TWh to 318 TWh, accounting for about 2.5% of the nation's total power usage. The report also notes that due to chip restrictions prompting some internet companies to shift computing power deployment to Southeast Asia, actual domestic data center power consumption in China may grow slower than the true pace of AI demand.
Three key factors are driving this substantial increase in power demand: the rapid growth of AIDC workloads; rising power consumption per chip as GPUs replace CPUs; and a significant expansion in system-level power requirements as rack density jumps from 10-15 kW to 100-120 kW and beyond.
AI Materials: Structural Supply Constraints
Regarding copper, Bank of America projects China's data center-related copper demand will grow from 341 kilotons in 2025 to 1,190 kilotons in 2030, a 28% compound annual growth rate. This would increase copper's share of China's total demand from 2.1% to 6.4%. The demand increment primarily comes from data center operations (650 kt), grid expansion (504 kt), and power plant construction (36 kt). Addressing market concerns about "fiber replacing copper," the report argues copper's position in power transmission, short-distance interconnects, and server internal connections is difficult to challenge. Coupled with a projected global copper supply deficit of 491-754 kilotons in 2026-2027, the rationale for supporting copper prices remains solid.
For PCB materials, while low-end copper foil capacity is structurally oversupplied, AI servers are driving increased PCB layer counts and higher specifications for high-frequency signal transmission, spurring explosive demand for high-end copper foil. High barriers to capacity conversion, equipment bottlenecks, and customer certification cycles typically exceeding one year make it difficult to close the supply-demand gap in the short term, supporting prices and profit margins for high-end products.
Supply of high-end electronic glass fiber (low Dk/low CTE specialty yarn) has historically been dominated by a few Japanese manufacturers like Nittobo. However, Chinese manufacturers have gradually achieved breakthroughs after years of R&D. Currently, there are five qualified suppliers in China, with Sinoma Science & Technology's Taishan Fiberglass holding a leading position. Bank of America assigns a buy rating to Sinoma Science & Technology, expecting its specialty glass fiber capacity to expand from the current 24 million meters to 94 million meters by 2027.
Optical fiber demand is accelerating its shift from traditional telecom to AIDC. While overall capacity is sufficient, the key upstream raw material for high-end fiber—preforms—faces long capacity expansion cycles and high technical barriers, creating a structural supply bottleneck that supports fiber prices. Bank of America gives a buy rating to Jiangsu Zhongtian Technology, forecasting its earnings per share to grow at a compound annual rate of about 75% from 2026 to 2027.
Regarding magnetic materials, tungsten, and uranium, high-performance neodymium iron boron (NdFeB) permanent magnets remain structurally tight, benefiting from dual drivers of AIDC liquid cooling systems and humanoid robot demand. Tungsten is entering the AI value chain due to expanding demand for high-density drilling in PCBs, with Chinese resource controls maintaining rigid supply. Uranium is viewed as a core strategic resource for scalable, zero-carbon, stable baseload power in the AI computing era. Bank of America's global commodities team predicts uranium prices will rise 47% and 29% year-on-year in 2026 and 2027, respectively, driven by an annual structural supply deficit of 2-7% and demand growing at a compound annual rate of about 4%.
AI Power Supply: China's Competitive Edge
Bank of America emphasizes that China possesses multiple structural advantages in the AI power supply sector: commercial electricity prices are 30-60% lower than in the US/EU; effective reserve capacity margin is about 30%, higher than the US's less than 25% and the EU's approximately 15%; the average age of transmission and distribution infrastructure is under 20 years (over 40 years in the US and Europe), offering higher network stability; and the country boasts 46 well-established ultra-high voltage transmission channels and a strong power equipment manufacturing supply chain. China has also announced plans to expand nuclear power installed capacity from 62GW at the end of 2025 to about 110GW by 2030 during its 15th Five-Year Plan period.
Opportunity 1: Transformers. Bank of America expects China's transformer exports to grow 30% year-on-year in 2026, with domestic grid investment increasing 12% to approximately 715 billion yuan. The global transformer shortage is expected to last at least until 2029, with high-voltage transformer lead times as long as three years. China's complete supply chain is effectively filling overseas capacity gaps.
Opportunity 2: Gas Turbines. The bank's global industrials team forecasts global gas turbine orders to average about 120GW annually from 2026 to 2028. Current new order lead times are 3-6 years, providing a window for Chinese manufacturers to compete on price and delivery time (as short as 13 months). Dongfang Electric is the only Chinese company with export capability for medium-to-large gas turbines. Its G50 model 50MW unit has completed sales of 10 units to a Canadian data center customer and has been exported to Kazakhstan and Indonesia; management plans to increase export capacity to 23 units by the end of 2027 and 45 units by the end of 2029.
Opportunity 3: Diesel Engines. Chinese manufacturers of large-bore diesel engines have obtained US UL and EPA certifications, allowing them to enter the North American AIDC backup power market. Bank of America estimates China's AIDC backup diesel engine demand will reach 8,500 units in 2026, with tight supply pressure expected to ease in 2027 as capacity expands.
Opportunity 4: Energy Storage Systems. The bank predicts the global battery energy storage system (BESS) market will grow at a compound annual rate of about 23% from 2025 to 2030, with AIDC-related BESS growing at about 27%. By 2030, global new AIDC BESS installations are projected to reach 70 GWh, accounting for roughly 8% of total global new installations.
Opportunity 5: Power Supply Systems. Bank of America forecasts the Chinese AIDC power supply system (UPS+HVDC+SST) market to grow at a compound annual rate of approximately 25% from 2025 to 2030. Nvidia is actively pushing its supply chain to transition to an 800VDC high-voltage direct current architecture along its hardware roadmap to cope with rising rack power density. The average selling price (ASP) of high-capacity power supply systems is significantly higher than traditional power units, with R&D barriers and customization characteristics forming strong competitive moats.
Liquid Cooling: Rapid Market Adoption
Liquid cooling is the fastest-growing sub-sector in China's data center cooling market. As rack power density continues to exceed the practical upper limit for air cooling (around 40 kW/rack), coupled with tightening domestic energy efficiency regulations, the adoption rate of liquid cooling is accelerating. Liquid cooling offers 20 to 50 times higher heat transfer efficiency than air cooling, potentially reducing Power Usage Effectiveness (PUE) to about 1.1.
Bank of America expects the penetration rate of liquid cooling in Chinese data centers to rise from 30% in 2025 to 70% in 2030. Liquid cooling demand is projected to expand from 1.4GW to 9.5GW, representing a compound annual growth rate of 47%. The overall data center cooling market size is forecast to grow to 70 billion yuan by 2030 from 2025, corresponding to an overall compound annual growth rate of 36%. Currently, cold plate liquid cooling holds over 90% market share, while immersion cooling's share is expected to increase from about 5% in 2025 to approximately 17% in 2030, representing a market size of 16 billion yuan.
Comments