Largest GPU clusters in the world

87 named GPU clusters: 61 running or partly running, 22 announced or under construction. Largest installed count on record: Colossus 1 with 220,000 GPUs, then Aurora (63,744) and El Capitan (46,080). Biggest build on paper: Meta Prometheus, targeting about 1,000,000 GPUs. Every number links to its source.

87 clusters 114 sources

Last verified

Colossus 1220,0003—xAI (SpaceXAI)NVIDIA H100, H200 and GB200operationalMemphis, Tennessee, United States300320242026-11-11
Aurora63,7441—Argonne Leadership Computing FacilityIntel Data Center GPU Max SeriesoperationalLemont, Illinois, United States—2025-012026-11-11
El Capitan46,0801—Lawrence Livermore National LaboratoryAMD Instinct MI300A APUoperationalLivermore, California, United States——2026-11-11
Frontier39,4241—Oak Ridge Leadership Computing FacilityAMD Instinct MI250XoperationalOak Ridge, Tennessee, United States——2026-11-11
Meta GenAI InfiniBand cluster24,5761—MetaNVIDIA H100operational———2026-11-11
Meta GenAI RoCE cluster24,5761—MetaNVIDIA H100operational———2026-11-11
Project Ceiba20,7361—AWSNVIDIA GB200 (Blackwell)operational———2026-11-11
Meta AI Research SuperCluster (RSC)16,0001—MetaNVIDIA A100operational———2026-11-11
Microsoft Azure Eagle14,4001—MicrosoftNVIDIA H100operational——2023-112026-11-11
Eni HPC713,9201—EniAMD Instinct MI300A APUoperationalFerrera Erbognone, Italy—2026-062026-11-11
Eni HPC613,8881—EniAMD Instinct MI250XoperationalFerrera Erbognone, Italy—2024-112026-11-11
Leonardo Booster13,8241—CINECANVIDIA A100-64 custom Ampere, 64 GiBoperationalBologna, Italy—2023-072026-11-11
ByteDance MegaScale Ampere cluster12,2881—ByteDanceNVIDIA Ampereoperational———2026-11-11
LUMI-G11,9122—LUMI consortiumAMD Instinct MI250XoperationalKajaani, Finland——2026-11-11
Alps Grace Hopper partition10,7521—CSCSNVIDIA GH200 (H100 GPU)operationalLugano and Lausanne, Switzerland—2024-092026-11-11
NVIDIA Eos (10,752-GPU system)10,7521—NVIDIANVIDIA H100operational———2026-11-11
CoreWeave Barcelona H200 supercomputer10,2241—CoreWeaveNVIDIA H200announcedBarcelona, Spain——2026-11-11
Yotta Shakti Cloud NM19,2161—Yotta Data ServicesNVIDIA H100 and L40SoperationalNavi Mumbai, India——2026-11-11
TensorWave MI325X training cluster8,1921—TensorWaveAMD Instinct MI325XoperationalNorth America (city not published)—2025-072026-11-11
Perlmutter GPU partition7,1681—NERSCNVIDIA A100 40GB and 80GBoperationalBerkeley, California, United States——2026-11-11
ABCI 3.06,1282—AIST Solutions (ABCI)NVIDIA H200 SXM5, 141 GB HBM3eoperationalKashiwa, Japan——2026-11-11
Isambard-AI5,4481—University of BristolNVIDIA GH200 Grace HopperoperationalBristol, United Kingdom—2025-072026-11-11
NVIDIA Eos (4,608-GPU system)4,6081—NVIDIANVIDIA H100operational———2026-11-11
Tuolumne4,6081—Lawrence Livermore National LaboratoryAMD Instinct MI300A APUoperationalLivermore, California, United States—20242026-11-11
MareNostrum 5 ACC4,4801—Barcelona Supercomputing CenterNVIDIA Hopper, 64 GB HBMoperationalBarcelona, Spain——2026-11-11
Horizon (TACC)4,0001—Texas Advanced Computing Center (UT Austin, NSF)NVIDIA GB200 (NVL4 nodes)partly operationalRound Rock, Texas, United States——2026-11-11
Nebius Modiin Blackwell cluster4,0001—NebiusNVIDIA HGX B200operationalModiin, Israel—2025-102026-11-11
JUWELS Booster3,7442—Jülich Supercomputing CentreNVIDIA A100 40 GBoperationalJülich, Germany——2026-11-11
Shaheen III GPU partition2,8001—KAUSTNVIDIA GH200 Grace HopperoperationalThuwal, Saudi Arabia——2026-11-11
Venado2,5601—Los Alamos National LaboratoryNVIDIA GH200 Grace HopperoperationalLos Alamos, New Mexico, United States——2026-11-11
NAVER Sejong2,2401—NAVERNVIDIA A100operationalSejong, South Korea——2026-11-11
Polaris2,2401—Argonne National LaboratoryNVIDIA A100operationalLemont, Illinois, United States——2026-11-11
DeepSeek-V3 H800 training cluster2,0481—DeepSeekNVIDIA H800————2026-11-11
Sakura Ishikari H100 deployment2,0161—Sakura InternetNVIDIA H100operationalIshikari, Japan—2024-082026-11-11
Yandex Chervonenkis1,5921—YandexNVIDIA A100 80GB—Sasovo, Ryazan Oblast, Russia——2026-11-11
Gefion1,5281—Danish Centre for AI Innovation (DCAI)NVIDIA H100operational——2024-102026-11-11
Jean Zay H100 extension1,4561—IDRIS / CNRS (GENCI)NVIDIA H100 80 GBoperationalSaclay, France——2026-11-11
Adastra GPU partition1,3521—CINESAMD Instinct MI250XoperationalMontpellier, France——2026-11-11
MareNostrum 5 AI training partition1,3441—Barcelona Supercomputing CenterNVIDIA GB200operationalBarcelona, Spain——2026-11-11
Yandex Lyapunov1,0961—YandexNVIDIA A100 40GB—Sasovo, Ryazan Oblast, Russia——2026-11-11
Yandex Galushkin1,0881—YandexNVIDIA A100 80GB—Vladimir, Vladimir Oblast, Russia——2026-11-11
Chan Zuckerberg Initiative biological AI cluster1,0241—Chan Zuckerberg InitiativeNVIDIA H100operational———2026-11-11
Scaleway Nabuchodonosor 20231,0161—ScalewayNVIDIA H100operationalSaint-Ouen-l'Aumône (Scaleway DC5, Paris region), France—2023-102026-11-11
TSUBAME4.09601—Institute of Science TokyoNVIDIA H100 SXM5operationalTokyo, Japan——2026-11-11
MeluXina accelerator module8002—LuxProvideNVIDIA A100 40GBoperationalBissen (LuxConnect DC2), Luxembourg——2026-11-11
Setonix AMD GPU partition7681—Pawsey Supercomputing CentreAMD Instinct MI250XoperationalKensington, Australia——2026-11-11
Berzelius Ampere7521—NAISSNVIDIA A100, 40GB or 80GBoperationalLinköping, Sweden——2026-11-11
MareNostrum 5 AI inference partition5761—Barcelona Supercomputing CenterNVIDIA GB300operationalBarcelona, Spain——2026-11-11
Recursion BioHive-25041—RecursionNVIDIA H100operationalSalt Lake City, Utah, United States—2024-052026-11-11
Alice Recoque——Jules Verne consortium (GENCI, CEA, SURF, GRNET) for EuroHPC JUAMD Instinct MI430X, 432 GB HBM4under constructionTGCC, CEA DAM Île-de-France, Bruyères-le-Châtel, France——2026-11-11
Argonne Equinox—10,0001Argonne National Laboratory (DOE)NVIDIA BlackwellannouncedLemont, Illinois, United States——2026-11-11
Argonne Solstice—100,0001Argonne National Laboratory (DOE)NVIDIA BlackwellannouncedLemont, Illinois, United States——2026-11-11
Colossus 2—440,0003xAI (SpaceXAI)NVIDIA GB200 and GB300operationalMemphis, Tennessee, United States94632025-102026-11-11
DeepL Mercury——DeepL SENVIDIA H100operationalFalun (EcoDataCenter 1), Dalarna, Sweden—20232026-11-11
Deutsche Telekom Industrial AI Cloud (Munich)—10,0001Deutsche TelekomNVIDIA B200 (DGX B200) and RTX PRO BlackwelloperationalTucherpark, Munich, Bavaria, Germany—20262026-11-11
Doudna (NERSC-10)——NERSC, Lawrence Berkeley National Laboratory (DOE)NVIDIA Vera RubinannouncedBerkeley, California, United States——2026-11-11
ExxonMobil Discovery 6——ExxonMobilNVIDIA GH200 Grace Hopperoperational——20252026-11-11
Foxconn Taiwan AI Factory (Big Innovation Company)—10,0001Big Innovation Company (Foxconn)NVIDIA Blackwell Ultra (GB300 NVL72)announcedSouthern Taiwan——2026-11-11
GMI Cloud Taiwan AI Factory—7,0001GMI CloudNVIDIA BlackwellannouncedTaoyuan, Taiwan——2026-11-11
HUMAIN GB300 AI factory (phase 1)—18,0001HUMAIN (Public Investment Fund)NVIDIA GB300announcedSaudi Arabia——2026-11-11
IBM Blue Vela——IBMNVIDIA H100operational———2026-11-11
IREN Horizon 1——IRENNVIDIA GB300 NVL72operationalChildress, Texas, United States5012026-082026-11-11
JUPITER Booster——Jülich Supercomputing Centre (EuroHPC JU)NVIDIA GH200operationalJülich, Germany—20252026-11-11
KISTI National Supercomputer No. 6—8,4961KISTI (Korea Institute of Science and Technology Information)NVIDIA GH200under construction———2026-11-11
Lambda Kansas City AI Factory——LambdaNVIDIA Blackwell UltraannouncedKansas City, Missouri, United States——2026-11-11
Lux (ORNL)—4,0001Oak Ridge National Laboratory (DOE)AMD Instinct MI355X, 288 GB HBM3Eunder constructionOak Ridge, Tennessee, United States——2026-11-11
Magic G4 on Google Cloud——Google Cloud (for Magic)NVIDIA H100announced———2026-11-11
Magic G5 on Google Cloud——Google Cloud (for Magic)NVIDIA Grace Blackwellannounced———2026-11-11
Meta Hyperion——Meta—under constructionRichland Parish, Louisiana, United States—20282026-11-11
Meta Prometheus—1,000,0001MetaNVIDIA GB200partly operationalNew Albany, Ohio, United States47122025-102026-11-11
Microsoft Fairwater Atlanta——MicrosoftNVIDIA GB200 NVL72operationalAtlanta area, Georgia, United States63622025-102026-11-11
Microsoft Fairwater Wisconsin——MicrosoftNVIDIA GPUspartly operationalMount Pleasant, Wisconsin, United States36922026-042026-11-11
Mistral AI Bruyères-le-Châtel cluster—13,8001Mistral AINVIDIA GB300announcedBruyères-le-Châtel (Eclairion campus), Île-de-France, France——2026-11-11
Nebius Kansas City GPU cluster——NebiusNVIDIA HopperannouncedKansas City, Missouri, United States——2026-11-11
Nebius Vineland GPU infrastructure——Nebius (tenant at DataOne)—partly operationalVineland, New Jersey, United States50320262026-11-11
Nscale Loughton AI campus—23,0401NscaleNVIDIA GB300announcedLoughton, Essex, United Kingdom——2026-11-11
Nscale Narvik (Stargate Norway)—52,0002NscaleNVIDIA GB300under constructionNarvik, Norway—20272026-11-11
Nscale Sines AI campus deployment——NscaleNVIDIA Blackwell UltraoperationalSines, Portugal33220262026-11-11
Nscale Texas Microsoft cluster—104,0001NscaleNVIDIA GB300announcedBarstow (Cedarvale site), Ward County, Texas, United States——2026-11-11
NVIDIA Israel-1—2,0481NVIDIANVIDIA H100 80GBpartly operationalIsrael—2023-112026-11-11
SoftBank Blackwell DGX SuperPOD——SoftBank Corp.NVIDIA B200operational——2025-072026-11-11
Stargate Abilene (Crusoe / Oracle)——Oracle (built by Crusoe, for OpenAI)NVIDIA GB200partly operationalAbilene, Texas, United States61832025-092026-11-11
Stargate UK (Nscale for OpenAI)—8,0001Nscale (OpenAI offtake)—announcedCobalt Park, North Tyneside, United Kingdom (one of several UK sites)——2026-11-11
Tesla Cortex 1——Tesla—operationalGigafactory Texas, Austin, Texas, United States1051—2026-11-11
Tesla Cortex 2——Tesla—operationalGigafactory Texas, Austin, Texas, United States1201—2026-11-11
Together AI–Hypertec Blackwell cluster—36,0001Together AI and Hypertec CloudNVIDIA GB200 NVL72announced———2026-11-11
Yotta Greater Noida D2 Blackwell deployment—30,0001YottaNVIDIA Blackwell B300 UltraannouncedGreater Noida, India——2026-11-11

Browse by kind

All lists

How this table is built

Full methodology

Every value comes from a public source: the owner’s engineering post, user documentation, filing or press release, or a named report. The small number next to a figure opens that source. An empty cell means the owner has not published the figure. Installed GPUs and planned GPUs sit in separate columns, so an announcement never inflates a count.

Each row carries the date we last checked it, and the newest check was 11 Nov 2026. Spot a stale number? Email hello@gpuclusterlist.com with the source and we update the row.

Questions buyers ask

What is the largest GPU cluster in the world?

By installed GPUs that a company has put on record, xAI’s Colossus 1 in Memphis leads with over 220,000 NVIDIA GPUs on more than 300 MW (Anthropic, 6 May 2026). SpaceX’s S-1 describes 440,000 GB200 and GB300 GPUs across Colossus 2’s clusters and next expansion, and Microsoft says its Fairwater Wisconsin campus holds hundreds of thousands of NVIDIA GPUs. By plan, Meta’s Prometheus in New Albany, Ohio targets about 1 million GPUs, with hundreds of thousands running today.

Which supercomputer has the most GPUs?

Argonne’s Aurora has the most of any public supercomputer in this list, with 63,744 Intel Data Center GPU Max GPUs. Lawrence Livermore’s El Capitan follows with 46,080 AMD MI300A accelerators and Oak Ridge’s Frontier with 39,424 AMD MI250X packages. JUPITER at Jülich runs around 24,000 NVIDIA GH200 superchips.

How many GPUs does xAI’s Colossus have?

Colossus 1 has over 220,000 NVIDIA GPUs and more than 300 MW, all of which Anthropic agreed in May 2026 to use. Tom’s Hardware reports a mix of 150,000 H100, 50,000 H200 and 30,000 GB200 for it. For Colossus 2, SpaceX’s S-1 describes about 110,000 GB200 and 110,000 GB300 in its first two clusters plus at least 220,000 more GB300, 440,000 in all, which the table carries as planned. Press reports of 550,000 are unconfirmed, and Epoch AI estimates 946 MW of IT load in service.

How do you tell an announced cluster from one that is running?

Every row carries one of four statuses. Operational means the operator or a primary source says it is running. Partly operational means a phase is live and more is coming, as at Stargate Abilene (6 of 8 buildings delivered). Under construction and announced cover clusters that are contracted or built but not yet confirmed live. Planned GPU counts sit in their own column and never add to the installed count.

How are GPUs counted, and why do some clusters show no count?

The installed count is physical GPUs: nodes times GPUs per node when the operator publishes both. MI250X boards count once, not as two dies, and an MI300A APU counts as one GPU. H100-equivalents and design limits are never counted as GPUs. When an owner publishes no installed total, as with Fairwater, Tesla Cortex and DeepL Mercury, the cell stays empty and the note column carries what the source does say.