Table of Contents
According to the NTT Global Data Centers report, securing sites and power capacity will not be enough if operators cannot obtain the equipment, workforce, and supporting infrastructure required to bring facilities online
In sum – what to know:
Three growth paths – NTT Global Data Centers models annual global data center IT capacity growth of 23%, 30% or about 15% through 2030, depending on AI demand and infrastructure constraints.
Capacity crunch risk – The report argues that power, grid connections, transformers, chips, land, labor, and capital are emerging simultaneously as limiting factors for AI infrastructure expansion.
Efficiency still matters – Even under the fastest growth scenario, improving cooling, hardware, and facility efficiency remains essential to reduce power overhead and support continued AI deployment.
Global AI infrastructure growth will depend less on demand than on the industry’s ability to overcome mounting constraints across power, supply chains, and construction, according to a new scenario analysis from NTT Global Data Centers and economic consultancy ThoughtLab.
In a recent report, the organizations modeled three possible trajectories for global data center expansion between 2026 and 2030. Depending on AI adoption, infrastructure availability, and regulatory conditions, global IT capacity could grow at an average annual rate of 30%, 23% or about 15%.
The report argues that AI demand itself is unlikely to be the limiting factor. Instead, it warns that the industry faces a growing convergence of bottlenecks spanning power availability, grid connections, transformers, semiconductor supply, land, capital, and skilled labor.
According to the analysis, securing sites and power capacity will not be enough if operators cannot obtain the equipment, workforce and supporting infrastructure required to bring facilities online.
“The AI era will reward those who treat data center growth as a system-wide challenge to be orchestrated, not a land grab to be won one project at a time,” the report states.
The baseline scenario assumes global data center IT capacity expands at a compound annual growth rate (CAGR) of 23% through 2030. Under this outlook, continued AI adoption and digitalization drive sustained expansion, while improvements in cooling technologies, IT hardware, and operational optimization gradually improve energy efficiency.
However, NTT expects this growth to face increasing pressure from stricter regulations, particularly in Europe, along with supply chain constraints affecting equipment such as gas turbines, servers and grid connections.
In the accelerated scenario, annual IT capacity growth reaches 30%, driven by stronger AI demand, hyperscale investment, and government support. Total global IT capacity increases from 57 GW in 2023 to 351.9 GW by 2030.
Under this scenario, operators accelerate construction of new AI facilities while retrofitting older sites with more efficient servers and cooling systems. The report projects Power Usage Effectiveness (PUE) for new AI data centers improving from 1.14 in 2026 to 1.085 by 2030, while new non-AI facilities improve from 1.30 to 1.25 over the same period.
The third scenario assumes infrastructure constraints become more severe. Regulatory hurdles, limited power availability, supply chain disruptions, and broader economic and geopolitical pressures reduce annual IT capacity growth to approximately 15%.
Slower expansion also reduces incentives to invest in next-generation efficiency technologies, resulting in more gradual improvements in PUE compared with the higher-growth scenarios, according to the report.
Across all three scenarios, NTT concludes that AI infrastructure growth will increasingly depend on coordinated planning across utilities, equipment manufacturers, investors, operators, and policymakers rather than on data center demand alone.
