The United States and China are leading the global build-out of data centers, driven by the intense competition for AI leadership. These two nations account for approximately 70% of global data center energy consumption. By 2030, worldwide installed data center capacity is projected to reach 220 gigawatts, a sixfold increase from 2020. This expansion is essential for training advanced AI models, which can cost hundreds of millions to over a billion dollars per run, and for continuous AI inference by millions of users.
Hyperscalers like Google, Amazon, Microsoft, and Meta are investing heavily, with plans to spend over $400 billion on data centers in 2026, building on over $350 billion in 2025. This massive investment has led to a building frenzy akin to 19th-century railway expansion, with some law firms reporting deals regularly exceeding $4 billion to $5 billion. JPMorgan predicts that over $5 trillion will be spent on AI infrastructure over the next five years, with McKinsey forecasting a 22% annual increase in data center demand.
The rapid expansion is straining existing infrastructure, particularly the US electrical grid, which S&P Global predicts will need 22% more power for data centers by the end of 2025 and nearly three times as much by 2030. Local communities are increasingly opposing new data center projects due to concerns about utility bills, power disruptions, high water consumption for cooling, constant noise, and limited local economic contributions. In the second quarter of 2025 alone, an estimated $98 billion in projects were blocked or delayed due to local action. Supply chain issues, equipment shortages, and permitting hurdles are also delaying construction, with up to 40% of 2026 projects at risk of falling behind schedule. Many experts believe nuclear power, including small modular reactors, will be a necessary long-term solution to meet the immense energy demands.
