ROOT / COMPUTE // AI TRAINING CLUSTERS
// Compute · decoded · deep-research pass

AI Training Clusters, decoded.

The gigawatt machines training frontier models, by GPU count and power draw.

SEMI-LIVE◊ confidence: Medium
6
mega-clusters
~1 GW
first crossed (Anthropic)
300 MW
top single site
2026
peak buildout
// the signal · what the data says
Frontier clusters are now measured in gigawatts, not servers. In March 2026 an Anthropic facility became the first to cross 1 GW, beating OpenAI's Stargate; xAI's Colossus in Memphis pioneered the ~12-month gigawatt buildout. Stargate targets 10 GW across five+ US sites. Cluster size is the clearest proxy for who can train the largest models.
// comparison · every entry, benchmarked & annotated
NameMakerGPUs (k)Power (MW)InterconnectOnline (yr)
Colossus / Colossus 2xAI200300NVLink+RoCE2025
◊ Memphis; fastest gigawatt-scale buildout on record
RainierAmazon/Anthropic160300EFA/Trainium22026
◊ Anthropic training on Trainium2; ~1 GW class
FairwaterMicrosoft150210NVLink2026
◊ Wisconsin; OpenAI + Microsoft
Stargate AbileneOpenAI/Oracle100300NVLink2026
◊ Most advanced Stargate site, ~0.3 GW live, scaling
TPU v7 podsGoogle120160Optical OCS2025
◊ Ironwood pods, 9,216 chips each
Meta HyperionMeta130190RoCE2026
◊ Louisiana; multi-GW plan on gas
// leaderboard · GPUs (k)
01Colossus / Colossus 2200
02Rainier160
03Fairwater150
04Meta Hyperion130
05TPU v7 pods120
06Stargate Abilene100
// sources & where to verify
// full index, per-field sourcing & CSV export   [ AUTHENTICATE → ]
◊ Deep-research pass, mid-2026. Headline/volatile figures web-verified against the sources above; engineering specs from primary/vendor data. Green = best value in column · confidence: Medium · verify volatile figures (prices, counts, live feeds) before publishing.
← AI Accelerators◊ indexHyperscaler Capex →