News
DealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory’s NVIDIA & AMD Enterprise ExpansionDealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory’s NVIDIA & AMD Enterprise Expansion
EveryNeocloud

Bare-metal GPUs

Whole servers, with no virtual layer. 85 providers on EveryNeocloud so far.

A bare-metal GPU server is a whole physical machine rented to one customer, with no hypervisor, the software layer that splits a server into virtual machines. The customer gets every GPU, all the memory and direct access to the network cards. Neoclouds such as CoreWeave built their business on bare metal, and hosts such as Hetzner, OVHcloud and Leaseweb have long rented dedicated servers.

Why AI teams want it

Large training jobs need every bit of speed, and a virtual layer can slow the links between GPUs and to the network. On bare metal, GPUs talk to each other through NVLink and to other servers through InfiniBand or RoCE, a fast Ethernet method, with nothing in between. Teams can also install their own drivers and tools. The trade-off is that the customer manages more of the system.

How it is sold

Bare-metal GPU servers are usually rented by the month or on reserved contracts, often as eight-GPU nodes. Many providers wrap bare metal in Kubernetes or a job scheduler such as Slurm, so customers get the speed of bare metal with tools to run jobs. Providers compete on how fast they can hand over healthy servers and how they handle failures.

Famous bare-metal GPUs providers

Still to be written: Leaseweb, Vultr.

Further reading

Oracle Cloud 🇺🇸 United States and 4 moreOracle Cloud Oracle's global cloud for servers, storage, databases and GPU computeHyperscaler · Other Nvidia GPUs · 2016 IREN 🇦🇺 Australia and 2 moreIREN Australian GPU cloud and data centre operator for AI workloadsNeocloud · Other Nvidia GPUs · 2018 Nebius Group 🇳🇱 Netherlands and 2 moreNebius Group AI infrastructure company offering GPU cloud services, listed on Nasdaq.Neocloud · Other Nvidia GPUs · 2004 Hyperion 🇺🇸 United StatesHyperion A planned AI data center in Louisiana owned by Meta Platforms.Hyperscaler 🇺🇸 United StatesCerebras Developer of wafer-scale AI chips and supercomputers for inference and training.Neocloud · Cerebras wafer-scale · 2016 Hetzner Online 🇩🇪 Germany and 3 moreHetzner Online German data centre operator selling servers, colocation and cloud capacityNeocloud · 1997 Colossus 🇺🇸 United StatesColossus An AI supercomputer cluster in Tennessee and Mississippi built by xAI.Supercomputer · Nvidia Hopper · 2024 Gcore 🇱🇺 Luxembourg and 1 moreGcore Luxembourg edge AI, cloud, CDN and security provider with GPU clusters in EuropeNeocloud · Other Nvidia GPUs · 2011 🇫🇷 France and 1 moreScaleway French cloud and dedicated server provider run by the Iliad groupNeocloud · 2000 SambaNova Systems 🇺🇸 United StatesSambaNova Systems American provider of custom AI silicon and cloud inference services based on Reconfigurable Dataflow Units.Neocloud · Other and custom chips · 2017 Nscale 🇬🇧 United KingdomNscale UK-based GPU cloud provider and data centre operator founded in 2024.Neocloud · Other Nvidia GPUs · 2024 Azure 🇺🇸 United StatesAzure Microsoft's global cloud computing platform for infrastructure and AI services.Hyperscaler · Nvidia Ampere · 2010 🇺🇸 United States and 2 moreAmazon Web Services Amazon's on-demand cloud platform, offering compute, storage and AI infrastructure worldwideHyperscaler · Other Nvidia GPUs · 2006 🇨🇭 SwitzerlandAlps Swiss national supercomputer using Nvidia GH200 chips for AI and scientific research.Supercomputer · Other Nvidia GPUs · 2024 🇺🇸 United States and 1 moreCoreWeave US GPU cloud running AI data centres for hyperscalers and model labsNeocloud · Nvidia Hopper · 2017 🇫🇷 FranceOVHcloud French cloud and hosting provider running its own servers and data centresNeocloud · 1999 Taiwania 🇹🇼 TaiwanTaiwania A series of supercomputers in Taiwan for scientific research and public access.Supercomputer · Other Nvidia GPUs · 2018 🇺🇸 United StatesNCAR-Wyoming Supercomputing Center A high-performance computing and data archival facility in Cheyenne, Wyoming.Supercomputer · Other Nvidia GPUs · 2003 Northern Data 🇩🇪 Germany and 4 moreNorthern Data German GPU cloud and data centre group for AI and high-performance computingNeocloud · Nvidia Hopper · 2019 🇷🇺 RussiaYandex Cloud Russian cloud computing and AI infrastructure providerNeocloud · Other Nvidia GPUs · 1997 IBM Cloud 🇺🇸 United States and 3 moreIBM Cloud IBM's hybrid cloud platform spanning data centres, bare metal and managed KubernetesHyperscaler 🇮🇹 Italy and 3 moreAruba Italian data centre and cloud provider running its own campuses across Italy and EuropeNeocloud · 1994 Ionos 🇩🇪 Germany and 2 moreIonos German web hosting and cloud computing provider, part of United InternetNeocloud · 1988 🇺🇸 United States and 1 moreAmazon Elastic Compute Cloud AWS service for renting virtual servers by the secondHyperscaler · Other Nvidia GPUs · 2006 Megaspeed 🇸🇬 Singapore and 4 moreMegaspeed Singapore AI cloud firm investigated over Nvidia chip diversion to ChinaNeocloud · Nvidia Blackwell · 2015 Verda 🇫🇮 FinlandVerda European full-stack AI cloud offering GPU clusters, instances and model APIsNeocloud · Nvidia Blackwell Novita AI 🇺🇸 United StatesNovita AI AI cloud platform offering model APIs and GPU instances for developers.Neocloud · Other Nvidia GPUs DeepInfra 🇺🇸 United StatesDeepInfra Inference cloud for open-source AI modelsInference platform · Nvidia Blackwell · 2022 Runware 🇺🇸 United StatesRunware Serverless AI inference platform with custom hardwareInference platform · Other Nvidia GPUs · 2023 Siam AI 🇹🇭 ThailandSiam AI GPU cloud provider based in Thailand for enterprise workloadsNeocloud · Other Nvidia GPUs RunPod 🇺🇸 United StatesRunPod AI developer cloud for training and inferenceNeocloud · Nvidia Blackwell Lightning AI 🇺🇸 United StatesLightning AI AI cloud platform with owned GPU infrastructureNeocloud · Other Nvidia GPUs · 2026 Exoscale 🇨🇭 Switzerland and 4 moreExoscale European cloud provider for developers and AI workloadsNeocloud · Other Nvidia GPUs · 2011 Zenlayer 🇺🇸 United StatesZenlayer Distributed cloud infrastructure for global AI inferenceNeocloud · 2014 Hostkey 🇳🇱 Netherlands and 3 moreHostkey Dedicated GPU and bare-metal server provider in Europe and the USNeocloud · Other Nvidia GPUs · 2007 Hivelocity 🇺🇸 United StatesHivelocity Bare metal and GPU infrastructure for AI workloadsNeocloud · Other Nvidia GPUs TensorDock 🇺🇸 United StatesTensorDock Marketplace for on-demand GPU servers and cloud instancesMarketplace · Nvidia Hopper Vast.ai 🇺🇸 United StatesVast.ai Marketplace for distributed GPU cloud computeMarketplace · Nvidia Blackwell · 2016 GPU Mart 🇺🇸 United StatesGPU Mart Dedicated GPU VPS and bare-metal servers for AI and rendering.Neocloud · Nvidia Blackwell · 2021 Hyperbolic 🇺🇸 United StatesHyperbolic Unified platform for global GPU compute access and orchestrationMarketplace · Nvidia Hopper Spheron Network WorldSpheron Network Marketplace for renting enterprise NVIDIA and AMD GPUsMarketplace · Nvidia Blackwell SaladCloud 🇺🇸 United StatesSaladCloud Distributed GPU cloud using idle consumer hardwareNeocloud · Other Nvidia GPUs · 2018 Clore.ai WorldClore.ai Decentralised marketplace for renting NVIDIA GPUsMarketplace · Other Nvidia GPUs Lium 🇰🇳 Saint Kitts and NevisLium GPU rental marketplace with per-second billingMarketplace · Nvidia Blackwell Phala Cloud 🇺🇸 United StatesPhala Cloud Verifiable private AI compute on GPU TEEsNeocloud · Nvidia Hopper Applied Digital 🇺🇸 United StatesApplied Digital Builder and operator of large-scale AI data centresNeocloud · 2021 DigiPowerX 🇺🇸 United StatesDigiPowerX Vertically integrated AI infrastructure with owned power and GPU compute.Neocloud · Nvidia Blackwell Bitzero 🇳🇴 Norway and 1 moreBitzero Nordic AI data centre operator with over 1 GW of power capacity.Neocloud

Photo credits on this page

The small pictures above come from Wikimedia Commons under free licences. Some are cropped to fit.

The weekly digest

Every Monday: the provider of the day, new pages on EveryNeocloud, and the conferences and anniversaries of the week ahead.