What is an AI datacenter?

An AI datacenter is a facility purpose-built or retrofitted for GPU and accelerator workloads — typically at rack densities above 30 kW, with liquid-cooling readiness and high-bandwidth fabric (InfiniBand NDR, NVLink, or Spectrum-X). At least 40% of IT load runs AI training or inference on GPUs such as NVIDIA H100, H200, B200, or GB200 NVL72.

Which datacenter has the most GPUs?

As of 2026, xAI Colossus in Memphis, Tennessee holds the title of the world's largest single-site GPU deployment, with more than 200,000 NVIDIA H100 and H200 GPUs in one InfiniBand domain. Meta Hyperion (Louisiana) and OpenAI Stargate Abilene (Texas) are scaling toward gigawatt-class deployments that will overtake it within the next 18 months.

What is the largest AI cluster?

xAI Colossus is the largest contiguous AI training cluster in 2026, exceeding 200,000 NVIDIA H100/H200 GPUs on a single InfiniBand NDR 400G fabric. Meta's Hyperion build and OpenAI's Stargate program are constructing multi-site fabrics that will exceed one million accelerators across coupled campuses by 2027.

Which GPU cloud is cheapest for H100?

On-demand: Together AI ($1.75/hr) and Voltage Park ($1.89/hr) lead. Reserved 1-year: Voltage Park ($1.20/hr) and Crusoe ($1.39/hr) are typically lowest. AWS p5 sits at the top of the price stack at ~$4.10/hr on-demand.

How is Datacenter.computer different from a datacenter directory?

Datacenter.computer is an agent-readable intelligence layer, not a directory. It normalizes pricing, capacity, power, fabric, and latency across 240+ facilities and 40+ operators into a single queryable graph for AI agents, traders, and procurement teams.

What is an AI factory?

An AI factory is a datacenter purpose-built for training and/or high-throughput inference — typically single-tenant, liquid-cooled, with 100kW+ racks and InfiniBand or NVLink fabrics across thousands of GPUs.

How is an AI factory different from a regular datacenter?

AI factories are optimized end-to-end for one workload class: dense GPU racks, dedicated substations, and topology-aware networking. Our Top 100 AI Datacenters list covers general-purpose AI capacity inside multi-tenant facilities.

What scale qualifies for the Top 100?

Sites must operate (or have firm commissioning dates) of at least 50 MW IT load dedicated to AI compute, or house clusters of 10,000+ frontier GPUs.

Are Stargate, Colossus, and Hyperion included?

Yes. Tier 1 covers flagship gigawatt-class factories including xAI Colossus (Memphis), OpenAI Stargate Abilene, Meta Hyperion (Louisiana), and Meta Prometheus (Ohio).

How does this list relate to GPU clusters?

A factory is the building; a GPU cluster is the system inside it. Our Top 100 GPU Clusters ranking decomposes these factories into discrete superclusters.

[Ranking 03/05] · 2026 Edition

Top 100 AI Factories

Dedicated AI factories — single-purpose training and inference sites built around NVIDIA, AMD, and custom-silicon clusters.

Top 100 Datacenters →Top 100 Operators →Top 100 GPU Clusters →Top 100 Infra Cos →Benchmarks →

[AI Summary]

The Top 100 AI Factories ranks single-purpose sites engineered exclusively for training and large-scale inference — not general cloud datacenters. The 2026 list covers xAI Colossus, OpenAI Stargate, Meta Hyperion and Prometheus, Microsoft and Oracle gigawatt builds, Crusoe and CoreWeave flagships, and sovereign factories in the UAE and Saudi Arabia, scored on contracted power, GPU density, and time to first training run.

[Cross-Reference]

Top 100 Datacenters →

Compare single-purpose AI factories with general AI datacenters in the broader facility ranking.

Top 100 GPU Clusters →

See the H100, H200, B200, and GB200 superclusters housed inside these factories.

Top 100 Operators →

Find the operators — hyperscalers, neoclouds, sovereigns — committing capital to AI factories.

Top 100 Infra Cos →

Trace the silicon, networking, and cooling vendors building inside these factories.

[T1]

Tier 1 — Frontier Lab Factories

Factories built directly for frontier model labs.

10 entries

001OpenAI Stargate Abilene
002Anthropic-Amazon New Carlisle
003xAI Colossus Memphis
004xAI Colossus 2 (TN)
005Meta Hyperion (LA)
006Meta Prometheus (OH)
007Microsoft Mt Pleasant AI Factory
008Microsoft Atlanta AI Factory
009Google Red Oak AI Factory
010Anthropic Project Rainier West

[T2]

Tier 2 — Hyperscaler AI Factories

Hyperscaler-owned AI training campuses.

10 entries

011AWS Project Rainier (IN)
012AWS Project Ceiba (TX)
013Azure AI Sydney
014Azure AI Mt Pleasant Phase II
015Google TPU v6 Council Bluffs
016Google TPU v6 St Ghislain
017Meta Eagle Mountain AI
018Meta Forest City AI
019Oracle Stargate I
020Oracle Stargate II

[T3]

Tier 3 — Neocloud AI Factories

Flagship factories from AI-native cloud providers.

10 entries

021CoreWeave Plano AI Factory
022CoreWeave Lancaster
023Crusoe Abilene AI Factory
024Crusoe Wyoming AI Factory
025Lambda Allen AI Factory
026Lambda Las Vegas
027Nebius Kansas City
028Nebius Mäntsälä
029Together AI Reno
030Voltage Park Houston

[T4]

Tier 4 — Sovereign AI Factories

Nation-funded AI factories.

10 entries

031G42 Stargate UAE
032Humain Riyadh AI Factory
033Humain Dammam AI Factory
034DataVolt NEOM AI Factory
035Saudi Aramco Dhahran AI
036G42 Condor Galaxy
037G42 Kenya AI Factory
038Bharti Airtel India AI Factory
039Tata AI Factory (Jamnagar)
040Reliance Jamnagar AI Factory

[T5]

Tier 5 — European AI Factories

EU AI Factory initiative and sovereign EU sites.

10 entries

041JUPITER AI Factory (Jülich, DE)
042Leonardo AI Factory (Bologna, IT)
043LUMI AI Factory (Kajaani, FI)
044MareNostrum AI Factory (Barcelona)
045MeluXina AI Factory (Luxembourg)
046Karolina AI Factory (Ostrava)
047Discoverer+ (Sofia)
048Deucalion (Minho, PT)
049Vega AI Factory (Maribor)
050Daedalus (Athens)

[T6]

Tier 6 — APAC AI Factories

Large APAC training factories.

10 entries

051Alibaba Zhangbei AI Factory
052Tencent Guian AI Factory
053ByteDance Tianjin AI Factory
054Baidu Yangquan AI Factory
055Huawei Ulanqab AI Factory
056NAVER Sejong AI Factory
057Kakao Ansan AI Factory
058SoftBank Hokkaido AI Factory
059NTT Inzai AI Factory
060SK Telecom Gasan AI Factory

[T7]

Tier 7 — Custom-Silicon Factories

Factories built around Trainium, TPU, MTIA, and other ASICs.

10 entries

061AWS Trainium2 Factory (IN)
062Google TPU v6 Factory (IA)
063Meta MTIA Factory (OH)
064Microsoft Maia Factory (WI)
065Cerebras G42 Wafer Factory
066Groq LPU Factory (Dallas)
067SambaNova SN40 Factory
068Tenstorrent Austin Factory
069Graphcore Bristol Factory
070Rebellions Seoul AI Factory

[T8]

Tier 8 — Inference-Optimized Factories

Factories tuned for high-throughput inference.

10 entries

071Groq Inference DFW
072Cerebras Inference Sunnyvale
073Together Inference Reno
074Fireworks Inference SJC
075Replicate Inference IAD
076Lambda Inference SJC
077RunPod Inference IAD
078DeepInfra LAX
079Modal Labs IAD
080OctoAI SEA

[T9]

Tier 9 — Emerging Independent Factories

Independent operators standing up dedicated factories.

10 entries

081Applied Digital Ellendale
082TeraWulf Lake Mariner AI
083Hut 8 Salt Creek AI
084Iris Energy Childress AI
085Core Scientific Denton AI
086Cipher Mining Black Pearl AI
087Marathon Granbury AI
088Galaxy Helios AI
089Riot Corsicana AI
090Cleanspark Wyoming AI

[T10]

Tier 10 — Next-Gen Gigawatt Factories

Announced gigawatt-class AI factories on the 2026–2028 roadmap.

10 entries

091Stargate Texas Phase II
092Stargate UAE Phase II
093Meta Louisiana Phase II
094AWS Indiana Phase II
095Microsoft Wisconsin Phase II
096Oracle Texas Gigafactory
097Crusoe Permian AI Factory
098CoreWeave Pennsylvania Mega
099Nscale Glomfjord AI Factory
100Helios West AI Factory (NV)

[FAQ]

Frequently asked questions

What is an AI factory?: An AI factory is a datacenter purpose-built for training and/or high-throughput inference — typically single-tenant, liquid-cooled, with 100kW+ racks and InfiniBand or NVLink fabrics across thousands of GPUs.
How is an AI factory different from a regular datacenter?: AI factories are optimized end-to-end for one workload class: dense GPU racks, dedicated substations, and topology-aware networking. Our Top 100 AI Datacenters list covers general-purpose AI capacity inside multi-tenant facilities.
What scale qualifies for the Top 100?: Sites must operate (or have firm commissioning dates) of at least 50 MW IT load dedicated to AI compute, or house clusters of 10,000+ frontier GPUs.
Are Stargate, Colossus, and Hyperion included?: Yes. Tier 1 covers flagship gigawatt-class factories including xAI Colossus (Memphis), OpenAI Stargate Abilene, Meta Hyperion (Louisiana), and Meta Prometheus (Ohio).
How does this list relate to GPU clusters?: A factory is the building; a GPU cluster is the system inside it. Our Top 100 GPU Clusters ranking decomposes these factories into discrete superclusters.

[Knowledge Graph]

Company → Datacenter → GPU → Customer → Industry

Every entity on this site is cross-linked. Follow the graph from operators down to specific facilities, GPU clusters, customers, and sectoral context.