China AI Infrastructure Spending to Triple by 2030
Power and cooling infrastructure emerge as critical bottlenecks as non-IT equipment now accounts for one-third of global AI capital expenditure.
China's artificial intelligence infrastructure spending is projected to triple by 2030, driven by aggressive investment in power and cooling systems that have emerged as critical constraints on AI deployment, according to Bank of America analyst Matty Zhao.
Beyond the chip shortage
While semiconductor supply has dominated AI infrastructure discussions, Zhao identifies power distribution and cooling capacity as the new limiting factors for large-scale AI buildouts. These systems now represent a substantial portion of total capital expenditure as data centers scale to accommodate power-hungry AI workloads.
Global AI capital expenditure is expected to increase fivefold by 2030, according to the Bank of America analysis first reported by CNBC. Notably, one-third of this spending now flows to non-IT infrastructure—the physical systems that deliver electricity and manage heat rather than the servers and chips that perform computations.
Equipment bottlenecks intensify
The infrastructure constraint has created severe supply chain pressure. Electrical transformers, essential components for stepping down high-voltage power to data center requirements, currently face lead times extending three years from order to delivery. This timeline forces AI companies and cloud providers to plan infrastructure investments far in advance of actual deployment.
The extended lead times reflect both surging demand and the specialized manufacturing requirements for high-capacity electrical equipment. Unlike semiconductors, where production can be ramped relatively quickly once fabrication capacity exists, power infrastructure requires heavy industrial manufacturing with limited global suppliers.
Why it matters
The shift in infrastructure bottlenecks from chips to power systems signals a maturation of AI deployment challenges. Companies racing to build AI capacity must now coordinate across electrical utilities, equipment manufacturers, and real estate developers—a far more complex undertaking than procuring servers. For China specifically, the tripling of AI infrastructure investment represents a strategic commitment to maintaining competitiveness in AI development despite ongoing technology export restrictions. The focus on power and cooling infrastructure also suggests Chinese planners anticipate sustained growth in domestic AI computing demand through the end of the decade.
The three-year transformer lead times create a natural brake on how quickly any region can scale AI infrastructure, regardless of available capital or chip supply. This timeline advantage may benefit early movers who secured equipment orders years ago.
The analysis was presented by Bank of America's Matty Zhao and reported by CNBC.
This is an original analysis by the Omega editorial team. Source reporting: AI Watch.
Want systems like this working for your business?
Book a Call
