GPU cluster lists
Six cuts of the same dataset, each answering one buyer question. The full table sorts and filters all 87 clusters.
- GPU supercomputers
National-lab and academic GPU supercomputers ranked by installed GPUs: Aurora, El Capitan, Frontier, JUPITER, LUMI and more, each count sourced and dated. 32 clusters - Big tech AI clusters
GPU clusters run by Meta, Microsoft, AWS, Tesla, NVIDIA, IBM and others: model, GPU count, location and power, with every figure linked to its source. 21 clusters - GPU cloud clusters
Named GPU clusters run by neoclouds and GPU clouds: GPU model, installed and contracted count, site, power and go-live status, each sourced and dated. 22 clusters - AI lab and enterprise clusters
GPU clusters owned by AI labs and enterprises: xAI Colossus 1 and 2, Mistral, DeepSeek, DeepL, ExxonMobil and more, with GPU model, count, power and status. 12 clusters - AMD GPU clusters
Every AMD Instinct cluster on the list, from El Capitan and Frontier to TensorWave’s 8,192-GPU MI325X cluster, ranked by GPU count with sources. 11 clusters - Announced GPU clusters
GPU clusters that are announced, contracted or under construction, with planned GPU counts, power and target dates, kept apart from installed capacity. 22 clusters