The rapid rise of artificial intelligence could overwhelm the United States' energy infrastructure by 2035, according to a new report by Deloitte. The report warns that power demand from AI-focused data centers is expected to surge more than 30 times, climbing from 4 gigawatts (GW) in 2024 to 123 GW within the next decade.
Deloitte’s analysis, based on its 2025 AI Infrastructure Survey, paints a stark picture of infrastructure readiness. It reveals widespread concern among executives in the power and data center industries: 72% cited grid and power capacity constraints as either “very” or “extremely” challenging for meeting anticipated demand.
Currently, AI accounts for 12% of U.S. data center energy consumption. By 2035, it is expected to account for 70% of an estimated 176 GW total power demand—an exponential increase fueled by widespread adoption of AI across sectors.
As investment in AI hardware—servers, chips, and supercomputers—accelerates, estimated to exceed $1 trillion over the next four years, power providers are under pressure to match pace. However, Deloitte notes a mismatch in funding and capacity building: while data centers enjoy robust private investment, utility companies face difficulties scaling their infrastructure without passing costs onto households.
Martin Stansbury, principal at Deloitte & Touche LLP, emphasized the urgency: “Infrastructure development offers a key opportunity to support AI and energy dominance, but the gaps we see today—technical, regulatory, and financial—require cross-sector collaboration to close.”
The report identifies three primary concerns among surveyed stakeholders: limited grid capacity, supply chain disruptions, and resource competition—particularly for energy. Nearly 80% of respondents believe AI will continue driving energy consumption higher, while 92% of data center operators view access to power capacity as a competitive challenge.
To overcome these hurdles, Deloitte recommends a dual approach: accelerating technological innovation—like advanced transformers, transmission systems, and renewable integration—and streamlining regulatory processes. It also proposes a new framework to reduce interconnection delays: expose the permitting process, expel speculative projects, and expedite priority developments.
Despite widespread agreement that cooperation between utilities and data centers is critical, confidence in current partnerships is low. Only 15% of data center respondents and 8% of power company executives described their collaboration as “highly effective.”
The report concludes that if the U.S. is to remain competitive in the global AI race, it must treat infrastructure development as a national strategic priority. Without immediate action to modernize power systems and build resilient, scalable data infrastructure, the digital economy may face a serious bottleneck.
Bd-pratidin English/ Jisan
