Neocloud Companies in 2026
Neoclouds are cloud providers that specialize in renting GPU compute for AI training and inference, rather than offering the broad catalog of services sold by hyperscalers such as AWS, Microsoft Azure and Google Cloud. They typically run their own large GPU fleets in purpose-built or leased data centers and fund expansion with multi-year capacity contracts from AI labs and hyperscalers, often backed by convertible notes and GPU-collateralized debt.
Neocloud Companies to know
We picked these companies for scale of contracted power and GPU fleet, size of their most recent financing or public-market status, named anchor customers (for example OpenAI, Microsoft, Meta, Anthropic), and independent press coverage as of October 2026. Every company runs its own GPU cloud and data-center capacity as its main business. Companies whose main identity is an inference API (Together AI), a GPU marketplace or serverless GPU provider (these go on the AI GPU Cloud page), or a hyperscaler were left out. The list is selected using our estimate of notability, which weighs scale, capital raised and contracted capacity.
CoreWeave
Nasdaq-listed GPU cloud provider running AI data centers for labs and hyperscalers
CoreWeave rents GPU cloud capacity built for AI training and inference, along with its own management software such as Mission Control. It reported Q1 2026 revenue of $2.08 billion, and a revenue backlog of $99.4 billion as of March 2026. Customers include OpenAI, Microsoft, Meta, Nvidia, Jane Street and, since April 2026, Anthropic.
- Headquarters
- Livingston, New Jersey, USA
Nebius
Amsterdam-based, Nasdaq-listed AI cloud running mostly company-owned data centers
Nebius Group runs an AI cloud on GPU clusters it hosts largely in its own data centers, including a flagship site in Finland. It emerged from the international businesses of Yandex N.V. after the Russian assets were sold in July 2024, and it also owns Avride and TripleTen. It has capacity deals with Microsoft and Meta, and reported Q2 2026 revenue of $582.3 million.
- Headquarters
- Amsterdam, Netherlands
Crusoe
Vertically integrated AI cloud and data-center developer focused on energy
Crusoe develops AI data centers and runs Crusoe Cloud, which offers GPU compute plus managed inference and fine-tuning. It started out turning flared natural gas into power for computing, and it builds OpenAI's 1.2 GW Stargate campus in Abilene, Texas. The company reports more than 6 GW of gross contracted capacity, about 1 GW of which is operational.
- Headquarters
- Denver, Colorado, USA
Nscale
London-based AI cloud building GPU data centers in Europe and the US
Nscale is a vertically integrated AI cloud that builds its own power, liquid-cooled data centers and GPU clusters for hyperscalers, AI labs and enterprises. Projects include Stargate Norway, a GPU site in Portugal and a Texas campus for Microsoft, and a deal to deploy up to 100,000 GPUs for Figure. It emerged from Australian crypto-mining firm Arkon Energy and was founded in 2024.
- Headquarters
- London, UK
Lambda
GPU cloud selling AI training and inference clusters
Lambda sells GPU cloud capacity, including on-demand instances, 1-Click Clusters and large dedicated superclusters for AI training and inference, and maintains the Lambda Stack deep learning software. It was founded in 2012 by machine learning engineers and has a large GPU supply deal with Microsoft.
- Headquarters
- San Jose, California, USA
Fluidstack
GPU cloud building dedicated AI clusters and data centers for frontier labs
Fluidstack runs single-tenant GPU clusters on Nvidia H100, H200, B200 and GB200 chips, using its Atlas OS for provisioning and Lighthouse for monitoring. Customers include Anthropic, Meta, Mistral and Poolside. It signed a $50 billion multi-year agreement to build Anthropic-dedicated data centers in Texas and New York. The company began in Oxford and is headquartered in London.
- Headquarters
- London, UK
IREN
Nasdaq-listed former bitcoin miner now running GPU AI cloud data centers
IREN runs AI Cloud services on its own renewable-powered data centers in North America and Australia and reported 150,000 GPUs as of March 2026. Contracts include a $9.7 billion agreement with Microsoft (November 2025) and a Nvidia partnership at its Childress, Texas site (May 2026). It reported FY26 revenue of $707.0 million, of which $128.8 million came from AI Cloud Services and most of the rest from bitcoin mining, and acquired Mirantis in August 2026.
- Headquarters
- Sydney, Australia
Firmus
Asia-Pacific AI factory and GPU cloud operator building Project Southgate
Firmus builds AI factories and runs Firmus AI Cloud, with sites in Singapore and Australia. Its Project Southgate is a rollout of AI factories across Australia, targeting 1.6 GW by 2028, and it has announced a development in Indonesia. The company reports about 900 MW of contracted capacity.
- Headquarters
- St Leonards, Australia
Vultr
Independent cloud provider with GPU-as-a-service across 30+ regions
Vultr is a privately held cloud provider offering compute, storage, bare metal and GPU-as-a-service. It is putting more emphasis on AI, including a planned investment of over $1 billion in a 50 MW AMD GPU cluster in Ohio. Unlike most neoclouds, it also sells general-purpose cloud services.
- Headquarters
- West Palm Beach, Florida, USA
TensorWave
AMD-only GPU cloud running Instinct accelerator clusters
TensorWave runs a GPU cloud built only on AMD Instinct accelerators, offering bare metal servers, managed Kubernetes and Slurm, and storage. Its investors include AMD, Magnetar and Atreides Management. Fermi America announced a $6.5 billion contract with TensorWave in August 2026.
- Headquarters
- Las Vegas, Nevada, USA
Neocloud Companies at a glance
| Company | Headquarters | Focus |
|---|---|---|
| CoreWeave | Livingston, New Jersey, USA | neocloud, gpu-cloud |
| Nebius | Amsterdam, Netherlands | neocloud, gpu-cloud |
| Crusoe | Denver, Colorado, USA | neocloud, gpu-cloud |
| Nscale | London, UK | neocloud, gpu-cloud |
| Lambda | San Jose, California, USA | neocloud, gpu-cloud |
| Fluidstack | London, UK | neocloud, gpu-cloud |
| IREN | Sydney, Australia | neocloud, gpu-cloud |
| Firmus | St Leonards, Australia | neocloud, gpu-cloud |
| Vultr | West Palm Beach, Florida, USA | neocloud, gpu-cloud |
| TensorWave | Las Vegas, Nevada, USA | neocloud, gpu-cloud |
Frequently asked questions
What is a neocloud?
A neocloud is a cloud provider that focuses mainly on renting GPU compute for AI training and inference. It does not offer the hundreds of services a hyperscaler like AWS, Azure or Google Cloud sells. Examples include CoreWeave, Nebius, Crusoe, Nscale and Lambda, all of which run large GPU fleets in their own or leased data centers.
Which neocloud companies are publicly traded?
Among notable neocloud companies, CoreWeave (Nasdaq: CRWV, IPO in March 2025), Nebius (Nasdaq: NBIS) and IREN (Nasdaq: IREN) are public. Nscale's September 2026 convertible notes convert to shares at its planned IPO, and Crusoe and Lambda have been reported to be preparing for possible IPOs.
Which neoclouds raised the largest rounds in 2026?
Large 2026 financings include Nebius's $5.75B convertible notes (August), Crusoe's $3.9B Series F at a $30.9B valuation (September), CoreWeave's $3.7B convertible notes (September), Nscale's $3.36B convertible notes led by Third Point (September), Firmus's $2B round at more than $10.5B (August), and a reported $1.5B Fluidstack round led by Jane Street at $18B (September).
How do neoclouds differ from hyperscalers?
Hyperscalers sell a broad catalog of services, including databases, storage and compliance tooling, across many regions. Neoclouds concentrate on GPU capacity, networking and the power to run it. Hyperscalers are also major neocloud customers: Microsoft has signed large contracts with CoreWeave, Nebius, Nscale, Lambda and IREN.
Who are the biggest customers of neoclouds?
AI labs, hyperscalers and trading firms make up much of the demand. Reported customers include OpenAI, Microsoft, Meta, Anthropic and Jane Street (CoreWeave); Microsoft and Meta (Nebius); OpenAI through the Stargate Abilene campus, plus Jane Street (Crusoe); Anthropic, Meta and Mistral (Fluidstack); Microsoft (IREN, Lambda and Nscale); and Figure (Nscale).
How is the Neocloud page different from the AI GPU Cloud page?
This page covers providers that run large GPU fleets and data centers of their own, often under multi-year contracts. The AI GPU Cloud page also includes GPU rental marketplaces and serverless GPU providers that may not own data-center capacity.
More AI company directories
- AI Inference CompaniesCerebras Systems, Fireworks AI, Groq and more
- Voice AI CompaniesElevenLabs, Deepgram, PolyAI and more
- AI Agent CompaniesSierra, Harvey, Decagon and more
- AI Infrastructure CompaniesDatabricks, Scale AI, Surge AI and more
- Healthcare AI CompaniesOpenEvidence, Abridge, Tempus AI and more
- AI Coding CompaniesCursor (Anysphere), Cognition, GitHub and more
About this directory: OpenCurious curates this list for researchers, founders, operators and buyers. Company descriptions and headquarters are based on publicly reported information as of and change often; figures a company reports about itself are its own claims. This is an editorial selection, not a paid ranking: no company paid to be listed, and companies appear in no particular order. Company names and logos are trademarks of their respective owners and are used only to identify each company. See our editorial policy. To suggest a company or a correction, email hello@opencurious.com.
The OpenCurious Newsletter
Stay curious. Get the best of the open web.
New directories, researcher tools, and open-source finds — straight to your inbox. No spam, ever.
- New directories
- Researcher tools
- Open-source finds
Occasional emails from OpenCurious. Unsubscribe anytime. Privacy Policy