Fluidstack

FluidStack runs AI cloud infrastructure at a scale that puts it in the center of some of the most sensitive compute environments on the planet. The company operates over 100,000 GPUs across its platform, serving AI labs, governments, and enterprises that can't afford downtime or data exposure. A $50 billion partnership with Anthropic and a €10 billion commitment to build a 1-gigawatt AI supercomputer in France signal the operational stakes - this isn't commodity cloud, it's infrastructure where a misconfigured bare-metal layer or a blind spot in monitoring could cascade across systems powering some of the largest AI models in deployment. The technical stack is built around two proprietary platforms. Atlas OS is a bare-metal operating system designed for fast provisioning and orchestration at scale - stripping away hypervisor overhead to give AI workloads direct hardware access. Lighthouse is an intelligent monitoring system built to catch anomalies and optimize workloads before they become incidents. Together they span GPU infrastructure, high-performance computing, and AI workload orchestration, with the kind of telemetry and control plane complexity that demands serious systems and security engineering. The company holds SOC 2 Type 2, ISO 27001, and GDPR certifications, with a stated focus on enterprise-grade security. Geographic footprint includes a global headquarters in New York City, the planned French supercomputer site, and additional custom data centers in the United States. For security practitioners, the threat surface is concrete: bare-metal operating systems managing GPU clusters at scale, monitoring pipelines carrying operational telemetry across jurisdictions, and workloads from high-profile clients in government and AI research. The attack vectors follow directly from that architecture.