Via nvidia.com
A single GB300 rack now pulls more power than most buildings, and the grid wasn't built for this
A single Nvidia GB300 rack reportedly draws over 180 kilowatts. For context, that’s roughly the electricity load of 60 average American homes, crammed into a server cabinet. Scale that up to a full AI cluster exceeding 100 megawatts, and you’re looking at something that rivals the electrical demand of a small city.
The numbers paint a stark picture #
US data centers consumed approximately 176 terawatt-hours (TWh) in 2025, accounting for about 4.4% of total national power generation. But projections suggest it could balloon to somewhere between 466 and 580 TWh by 2030, pushing data centers to consume 9-12% of all US electricity generation.
Goldman Sachs forecasts US data center power demand climbing from 31 gigawatts in 2025 to 66 gigawatts by 2027. That’s more than a doubling in just two years.
The supply side looks far less ambitious. Utilities can realistically deliver about 93 GW of additional practical power supply, leaving a substantial shortfall relative to AI data center demand.
The bottleneck isn’t chips anymore #
A new constraint has jumped the queue: physical infrastructure. Transformers, switchgear, and other grid equipment are now in critically short supply. Without them, even the most advanced GPU cluster sits dark.
Manufacturing lead times for large power transformers have stretched significantly, creating a cascading delay effect across the data center pipeline. You can order the servers, lease the land, and sign the contracts, but if the transformer won’t arrive for 18 months, none of it matters.
This dynamic hands a major advantage to two groups: utilities that happen to have spare generation capacity and the original equipment manufacturers (OEMs) that build grid infrastructure.
Political scrutiny and competing demands #
In some regions, data center power requests have been denied or delayed because local grids simply can’t absorb the load without risking reliability for existing customers.
For crypto miners, particularly those running large-scale proof-of-work operations, this competition is familiar territory. Bitcoin miners have spent years navigating power procurement, negotiating with utilities, and relocating to regions with cheap, abundant electricity. Some of those miners now sit on valuable power purchase agreements and grid interconnections that AI companies covet. The convergence of AI and crypto mining around power access has already spawned deals where mining companies pivot or lease capacity to AI workloads.
Who actually benefits #
Utilities with available capacity can command premium rates and favorable contract terms. Power equipment manufacturers face a demand surge that could sustain elevated revenues for years. Independent power producers, particularly those with generation assets near major data center corridors, hold increasingly valuable positions.
Nvidia itself occupies an interesting spot. Its chips are driving the demand explosion, but if energy constraints slow data center buildouts, that’s fewer GPU orders. Nvidia needs the grid to keep up with its product roadmap, and right now, the grid is falling behind.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our