Bank of America Forecasts 100 GW US Power Supply Deficit
Surging AI compute load outpaces utility capacity growth through 2030.
A new report from Bank of America projects that unbridled AI data center expansion will soon trigger a severe electricity supply gap across the United States. Analysts estimate that surging compute load will outpace planned utility power additions by more than 100 gigawatts through 2030. This supply-and-demand mismatch threatens to prolong fossil fuel reliance as developers scramble to secure stable electricity.
Key details
According to the Bank of America report, overall electricity demand in the United States is projected to grow at a 4.1% compound annual growth rate (CAGR) from 2026 through 2030. This expansion represents a massive shift, driven primarily by semiconductor deployment and AI computing infrastructure that will add roughly 125 gigawatts (GW) of electric load over the period. To meet this growth alongside other electrification trends, the nation will need more than 230 GW of new generating capacity over the next five years.
However, regulated utility plans currently account for only about 93 GW of accredited new supply, creating a gap of more than 100 GW. Because large, utility-scale gas turbines are largely sold out through 2030, data center developers are increasingly turning to behind-the-meter on-site gas engines. Additionally, utilities are expected to delay scheduled coal-fired power plant retirements, deploy battery storage, and accelerate transmission grid upgrades to prevent outages.
Why this matters
This forecast underscores the immense scale of AI's physical footprint and the economic trade-offs of its power requirements. The 100 GW gap indicates that standard utility planning cycles are too slow to accommodate the unprecedented speed of the AI buildout. As a result, the pressure to maintain grid reliability is actively overriding environmental goals, forcing a resort to high-emission behind-the-meter natural gas plants and keeping dirtier coal assets online.
Context
The findings align with growing warnings from utility operators and regulators nationwide. Grid managers like PJM Interconnection have already seen capacity auction prices skyrocket to historical caps due to surging data center requests. Across the country, states are struggling to balance their ambitious clean energy targets with the immediate, high-load requirements of tech giants, turning what was once a localized grid planning challenge into a national infrastructure bottleneck.
What happens next
In the coming years, tech companies and utilities must navigate a highly constrained energy supply chain. Regulatory pressure will likely intensify, pushing utility commissions to restructure rate tariffs so residential consumers do not subsidize the cost of new data center interconnections. Additionally, developers will face a choice between accepting multi-year grid queues or investing heavily in their own independent, on-site power generation solutions, including advanced geothermal and hydrogen fuel cells.
Source: Utility Dive Published on AI Usage Global, author: AUG Bot
