Fluidstack

Fluidstack is an AI cloud platform that builds and operates massive-scale, single-tenant GPU infrastructure for AI labs, governments, and enterprises training and deploying AI models. The company manages over 100,000 GPUs and supplies high-performance clusters built on NVIDIA H100, H200, B200 and GB200 systems, isolated at the hardware, network and storage levels with no shared resources or noisy neighbours.

Its technology stack spans bare-metal operating systems, cluster provisioning and orchestration, workload monitoring and optimisation, and hardware-level security. Two in-house products anchor the platform: Atlas OS, a bare-metal operating system designed to provision and orchestrate AI infrastructure quickly at scale, and Lighthouse, a monitoring system that optimises workloads and detects problems before they affect operations. Named partners include Anthropic, Mistral, Character.AI, Poolside and Black Forest Labs.

Fluidstack is expanding on both sides of the Atlantic. In 2025 it announced a partnership with Anthropic worth $50 billion to build custom data centres across the United States, and has committed €10 billion to a 1-gigawatt AI supercomputer in France. The company relocated its global headquarters to New York City in 2025 and operates internationally.

Security and isolation are central to the offering. Fluidstack holds SOC 2 Type 2 and ISO 27001 certifications and is GDPR compliant. Technical work covers GPU infrastructure, high-performance computing, AI model training and inference, bare-metal systems engineering, cluster orchestration, workload optimisation, and hardware, network and storage isolation.