Global data center spending is expected to more than double over the next two years as demand for AI services drives investment in computing infrastructure. According to BofA Global Research, spending could reach USD 1 trillion in 2026 before climbing to USD 2.1 trillion by 2028.
The forecast shows how quickly companies are building infrastructure to support the development and deployment of AI algorithms. Data centers need many servers, powerful hardware, robust power supplies, and effective cooling technology to handle increasingly demanding computing tasks.
AI Demand Drives Data Center Expansion
The data center market is growing rapidly because of investments in AI services and the computing power they require. The market, valued at USD 275 billion in 2022, is expected to reach USD 2.93 trillion in 2030, according to BofA Securities.
The market growth rate is expected to be 37% per annum between 2025 and 2030. The USA is becoming a leader in developing AI infrastructure, where companies producing chips and servers get more investment.
This growth also creates opportunities for suppliers of industrial equipment, including generators, switchgear, construction machinery, and cooling units.
Servers Account for the Biggest Costs
Building data centers requires substantial investment, particularly in computing hardware. BofA estimates that construction costs average USD 49 million per megawatt of capacity, with IT-related expenses accounting for 78% of total spending.
Servers are the largest cost component, estimated at USD 31.4 million per megawatt. Networking equipment follows at approximately USD 4.8 million per megawatt.
Other essential systems include power distribution equipment, backup generators, chillers, and cooling infrastructure. Depending on the function, these components can cost between USD 100,000 and USD 1.1 million per megawatt, according to the brokerage.
Also Read: Anthropic Secures First Canadian Data Centre Deal with EverWind
AI-Focused Data Centers Add to Investment
AI-focused data centers, sometimes called neoclouds, are also contributing to the investment surge. BofA estimates that these facilities cost around USD 57 million per megawatt, higher than the average construction cost for data centers overall.
However, AI-focused facilities can require less physical space than traditional data centers, even as they demand substantial computing power and specialised infrastructure.
As AI adoption spreads, investment is likely to extend beyond software developers and chipmakers to the companies supplying the equipment and systems that keep data centers running. The scale of projected spending suggests infrastructure will remain central to the AI industry's growth in the coming years.