Fluidstack
6.3/10Fluidstack builds and runs large GPU clusters and custom AI data centers for frontier AI labs and enterprises that train and serve large models.
Data last updated November 2026 · Our methodology
Overview
Fluidstack is an AI infrastructure company that started as a marketplace for spare GPU capacity. It now builds and operates large dedicated GPU clusters and purpose-built data centers. Customers get bare-metal capacity with managed Slurm or Kubernetes on top, plus active and passive cluster health checks and observability. Fluidstack names Anthropic, NVIDIA, Meta and Mistral AI among the companies it works with. In November 2025 Anthropic picked it to build gigawatt-scale data centers in Texas and New York, with sites coming online through 2026. SemiAnalysis gave its clusters a ClusterMAX Gold rating. The company raised an $830M Series A in January 2026 and is moving its global headquarters to New York City. Pricing is negotiated. Fluidstack's site has no public rate card, and capacity is sold through reserved cluster contracts and multi-year private cloud deals. Third-party trackers have quoted on-demand H100 rates of roughly $2.10 to $4.50 per hour, but those figures conflict with each other and may be out of date. Fluidstack suits AI labs and large enterprises that need thousands of GPUs with dedicated engineering support. It does not fit small teams looking for self-serve instances, web hosting or managed application services. Its main limitation is access: the best configurations require a sales conversation, and there is little public documentation on APIs, storage or security certifications.
Our Rating
Hosting & Infrastructure6.6/10
AI Software5.9/10
Who Fluidstack is best for
Good fit
Fluidstack sells multi-year dedicated clusters and custom data centers to very large AI buyers such as Anthropic.
Less suited
Well-funded AI startups like Mistral use it, but having to go through sales and contracts makes it a poor fit for small experiments.
There is no self-serve signup and little public documentation, so individual developers will find it hard to use.
Pros & Cons
Pros
- Can build and deliver capacity at gigawatt scale, which is why Anthropic chose it for its Texas and New York data centers.
- Runs managed Slurm and Kubernetes on bare metal, so training teams do not have to operate the cluster themselves.
- SemiAnalysis gave its GPU clusters a ClusterMAX Gold rating.
- Heavily funded, including an $830M Series A in January 2026, which lowers the risk that the supplier fails.
Cons
- There is no public rate card or self-serve signup, so pricing requires a sales quote.
- Public documentation on APIs, storage and compliance certifications is scarce.
- Third-party sources say its GPU catalog is narrower than marketplace providers and that on-demand billing is per hour.
- It offers no general web hosting, managed databases or app platform services.
Key Features
Not included
- Managed WordPress
- Shared Hosting
Fluidstack Pricing
Reserved and private cloud clusters
- Bare-metal GPU clusters
- Contact sales for a quote
- Custom data center builds
- Managed Slurm or Kubernetes
- Cluster health checks and observability
Frequently Asked Questions
No. Its site has no public rate card, and capacity is quoted per contract. Third-party trackers list H100 rates of roughly $2.10 to $4.50 per hour, but those figures conflict and may be out of date.
Current listings say there is no online signup. Capacity is contracted directly through sales.
It provides managed Slurm and Kubernetes clusters on bare-metal GPU servers, with active and passive health checks.
Fluidstack names Anthropic, NVIDIA and Meta, and press coverage names Mistral AI. Anthropic picked Fluidstack to build data centers in Texas and New York as part of a $50 billion plan.
Listings show locations in Canada, the Netherlands, Norway and Sweden. Custom data centers are being built in Texas and New York, and the company's headquarters is moving to New York City.
Similar Tools
See all alternatives
Amazon's public cloud platform, with over 200 pay-as-you-go services for compute, storage, databases, networking and AI, used by startups through large enterprises and governments.
Google's public cloud platform providing pay-as-you-go virtual machines, Kubernetes, serverless, databases, storage, networking, data analytics and AI services for businesses of every size.
Northflank is a Kubernetes-based developer platform for deploying services, jobs, databases and GPU workloads on its managed cloud or on customers' own cloud accounts.