U.S. Electric Grid Can't Keep Pace with AI Data Center Buildout

The infrastructure gap

America's electric grid is struggling to keep up with the breakneck expansion of AI data centers, creating a fundamental bottleneck that could delay more than half of planned facilities and force others to accept unreliable power connections.

OpenAI, Google, and Meta can finance and construct AI infrastructure in one to two years, but the utilities that must power these facilities operate on entirely different timelines. Building the generation capacity and transmission lines to deliver a gigawatt of electricity—enough to run a major data center—typically requires five years from initial grid connection request to commercial operation, according to research from Lawrence Berkeley National Lab.

"There is a real disconnect there," Rob Gramlich, president of consulting firm Grid Strategies, told Fortune. Tech companies move fast while utilities "notoriously move very slowly" because coordinating thousands of interconnected grid components demands "deliberate study and planning" that spans years.

Why it matters

This timing mismatch threatens the AI buildout that tech companies and investors are betting on. Without adequate power, ambitious deployment schedules collapse. The constraint also means utilities must make massive infrastructure investments based on data center commitments that may not materialize, creating financial risk that's now being pushed back onto developers through higher upfront costs.

Demand outpacing decades of precedent

AI data centers are projected to consume nearly 12% of all U.S. electricity by 2030—almost six times their 2018 share before the AI boom, according to Berkeley Lab. Grid watchdog NERC's 2025 reliability assessment shows summer peak electricity demand in North America will grow by more than 224 gigawatts over the next decade, 69% above the previous year's projection, driven primarily by AI facilities.

In the Western U.S. grid region, planned data centers average 10% of demand forecasts and reach as high as 40% in some areas.

The grid entered this surge already constrained. After big expansions in the 1980s and 1990s created excess capacity, the U.S. electricity industry experienced 25 years of flat demand and "got out of practice on building new infrastructure," Gramlich said. Between 2000 and 2023, electricity consumption grew far more slowly than in previous decades.

Now multiple forces are converging simultaneously: electric vehicles, electric heating, advanced manufacturing, and data centers. "The growth is about half data centers and half those other new things," Gramlich noted, "but those new quickly expanding uses of electricity are happening all at once."

The bottleneck effect

Kathryn Burke, who leads U.S. specialty energy and power growth at insurance firm Marsh, identified power access as "probably the number one, if not top five bottlenecks for data center development in the U.S. right now."

Companies are requesting more power at faster delivery rates than the grid can handle. Because utilities must invest in new generation capacity and grid upgrades to serve data centers, they're now demanding more money upfront from developers to avoid being "stranded" if projects fail to materialize.

The most immediate consequence likely won't be widespread blackouts. Utilities typically refuse to connect new customers they cannot reliably serve. Instead, data centers may wait years for full service or accept provisional connections where their power can be interrupted during grid strain.

"There is scarcity on the grid," Gramlich said. "Not everybody's going to get the full level of service that they want, at least until the system can catch up to these new demands."

Burke predicts 50 to 60% of data center projects will face delays and won't become operational within companies' hoped-for one-to-two-year timeframe. The Department of Energy acknowledged in July a "pressing need" for more infrastructure "due to load growth from data centers" among other factors.

These details were first reported by Fortune.