Training clusters
Reserved clusters of thousands of GPUs on one fabric. 105 providers on EveryNeocloud so far.
A training cluster is a block of hundreds to hundreds of thousands of GPUs joined on one fast network, reserved by one customer for months or years. This is how frontier models are trained, and it is the core business of the largest neoclouds. Deals are signed as contracts worth billions of dollars, often before the data centre is built.
What makes a cluster
Inside each server, GPUs talk over NVLink; between servers they use InfiniBand or fast Ethernet in a layout called a fat tree, so any GPU can reach any other at full speed. Shared storage feeds training data, and software such as Slurm or Kubernetes schedules jobs and restarts them when a GPU fails, which in a cluster of 100,000 GPUs happens many times a day.
How it is sold
Customers sign take-or-pay contracts: they pay for the capacity whether or not they use it. Microsoft, OpenAI, Meta and Google have signed such deals with CoreWeave, Nebius, Nscale, IREN and others. The contracts let providers borrow against them to buy GPUs. Critics say this ties a small number of buyers and lenders together; providers say long contracts make the business stable.
Famous training clusters providers
Further reading
馃嚭馃嚫 United StatesNCAR-Wyoming Supercomputing Center
A high-performance computing and data archival facility in Cheyenne, Wyoming.Supercomputer 路 Other Nvidia GPUs 路 2003
馃嚬馃嚰 TaiwanTaiwania
A series of supercomputers in Taiwan for scientific research and public access.Supercomputer 路 Other Nvidia GPUs 路 2018
馃嚥馃嚲 Malaysia and 4 moreYTL AI Cloud
Malaysian AI cloud and supercomputer run from a solar-backed Johor data centre parkNeocloud 路 Nvidia Blackwell 路 2024
馃嚚馃嚟 SwitzerlandSwiss National Supercomputing Centre
Switzerland's national high-performance computing centre, run by ETH Zurich in Lugano.Supercomputer 路 1992
馃嚭馃嚫 United StatesColossus
An AI supercomputer cluster in Tennessee and Mississippi built by xAI.Supercomputer 路 Nvidia Hopper 路 2024
馃嚠馃嚦 IndiaCentre for Development of Advanced Computing
Indian government society building supercomputers and advanced computing systemsSupercomputer 路 Other and custom chips 路 1988
馃嚬馃嚰 TaiwanTaiwania 3
A CPU-based supercomputer for scientific research and biomedical analysis in Taiwan.Supercomputer 路 Other Nvidia GPUs 路 2020
馃嚭馃嚫 United StatesPerlmutter
A DOE supercomputer at NERSC for fusion and climate research.Supercomputer 路 Nvidia Ampere 路 2021
馃嚚馃嚟 SwitzerlandAlps
Swiss national supercomputer using Nvidia GH200 chips for AI and scientific research.Supercomputer 路 Other Nvidia GPUs 路 2024
馃嚭馃嚫 United StatesStarcloud
Orbital data centres built and operated by a US space computing companyNeocloud 路 Nvidia Hopper 路 2023
Photo credits on this page
The small pictures above come from Wikimedia Commons under free licences. Some are cropped to fit.
- CoreWeave: Wikideas1, CC0, source
- JUWELS: Forschungszentrum J眉lich, CC BY-SA 4.0, source
- NCAR-Wyoming Supercomputing Center: National Science Foundation, Public domain, source
- Barcelona Supercomputing Center: Herodotptlomeu, CC BY-SA 4.0, source
- CINECA: Carlo Dani, CC BY-SA 4.0, source
- Swiss National Supercomputing Centre: CSCS, CC BY-SA 3.0, source
- SoftBank Group: Akonnchiroll, CC BY-SA 4.0, source
- Alps: Marco Abram, CSCS, CC BY-SA 3.0, source
- Galileo: CINECA, CC BY 2.0, source
- Spanish Supercomputing Network: David Abi谩n, CC BY-SA 4.0, source
- National Center for Supercomputing Applications: en:User:Ragib / Site , User:Ragib, CC BY 2.5, source
- Texas Advanced Computing Center: UNM Research Data Services, CC BY-SA 2.0, source
- National Computational Infrastructure: Nick-D, CC BY-SA 3.0, source
- YTL AI Cloud: YTL Power International Berhad, CC BY-SA 4.0, source
- Google data centers: erwinboogert, CC BY-SA 3.0, source