Computing · Projected milestone

Gigawatt AI campuses

Estimated year~2027

75% by 2029 · including earlier years

The year is an editorial timing estimate. The percentage applies to occurrence by the stated deadline. These estimates are not calibrated odds.

Single AI training sites draw over a gigawatt of power, the output of a large nuclear reactor.

Explore this milestone on the interactive map

What the forecast means

Milestone stage: deployment

plausible-unconfirmed

A built and energized single-site gigawatt AI campus, as distinct from announced plans.

Completion criteria: Operator and utility or independently audited metering show a sustained 24-hour average above 1 GW at the single campus. Include cooling and support load; proposed capacity, aggregate global sites, short spikes and reserved grid connections do not qualify.

This page describes a forecast, rather than an achieved result. Read the methodology and assumptions.

Built on

National electricity gridRecorded · 1926enables
AI supercomputer campusesRecorded · 2024enables
Liquid-cooled AI racksRecorded · 2024enables

Led to

AI research run by agent swarmsForecast · ~2028enables
Expert-level general AIForecast · ~2028enables
Million-agent swarmsForecast · ~2030influences
Zettascale supercomputersForecast · ~2035enables

Sources & evidence

Key Questions on Energy and AI: Executive summary 2026-04-16
The IEA identifies power, grid connection, memory/chip supply and financing bottlenecks alongside rapid efficiency gains. Announced data-center capacity and actual operational demand are different quantities.

Record reviewed 2026-10-18