AI Data Centers by Capacity
Capacity tier is one of the clearest ways to understand the AI infrastructure buildout. Browse 344 tracked facilities by disclosed megawatt scale, from sub-100 MW clusters to multi-gigawatt campuses.
Under 100 MW
Smaller AI clusters, sovereign systems, and early-stage regional builds.
99
facilities
4.1 GW
100-500 MW
Large regional campuses and hyperscale AI expansions.
146
facilities
30.4 GW
500 MW to 1 GW
Megacampuses pushing into utility-scale power demand.
44
facilities
26.4 GW
Over 1 GW
The world’s biggest AI infrastructure projects and multi-phase campuses.
46
facilities
94.7 GW
Capacity Unknown
Facilities with no public megawatt disclosure yet.
9
facilities
What These Tiers Mean
Capacity reflects publicly disclosed power demand, usually in megawatts or gigawatts. It is a lower bound on scale, not a perfect proxy for GPU count. A 50 MW sovereign AI cluster, a 300 MW hyperscale expansion, and a 1+ GW megacampus represent fundamentally different infrastructure classes. The unknown bucket is also useful: it shows where operators are announcing projects without disclosing the size of the buildout.