Epoch AI data centers
Citation
Epoch AI. “AI data centers.” Data explorer, no date (cites works to June 2026). https://epoch.ai/data/ai-data-centers. Read 2026-10-28.
Bottom line
- Epoch AI tracks “96 sites” of the world’s largest AI data centers, with estimates of power, compute and cost.
- No California site is in the list on the page. The largest sites are in Texas, Ohio, Virginia, Nebraska and Iowa.
Background
The page says its estimates are “built from high-resolution satellite imagery, permits, and our understanding of how data centers are constructed”. It claims “44% coverage”.
Key claims
- Totals: “96 sites”, “15.2M H100-equivalents”, “14.5 GW IT power”, “44% coverage” (header).
- Location: “The United States has the most large-scale AI data centers, concentrated in Texas (14 sites covered), Ohio (7 sites covered), Virginia (7 sites covered), Nebraska (4 sites covered), and Iowa (4 sites covered)” (FAQ, “Where are AI data centers located?”). The page names no California site.
- Largest by power: “Anthropic-Amazon New Carlisle, owned by Amazon in New Carlisle, IN, drawing 1,037 MW” (FAQ).
- Largest by compute: “Colossus 2, owned by SpaceXAI in Memphis, TN, with 1,112k H100e” (FAQ).
- Overhead: “Total facility power is 20-50% higher due to cooling and other infrastructure overhead”. Average use is “typically 60–80% of capacity” (FAQ, “How much power do AI data centers consume?”).
- Hardware: “NVIDIA H100, H200, and B200 GPUs, Google TPU v5 and v6 chips, and AWS Trainium2 accelerators” (FAQ).
Evidence and method
A research nonprofit’s estimates from imagery and permits, with each data point “vetted by our team”.
Limits
- It covers only the largest sites, about 44 percent by its own measure. University and state clusters in California are not in its scope.
- The figures are estimates.
- The page copy lists only the first sites of the directory.