Fluidstack
Category: AI Infrastructure
A specialized AI cloud platform providing rapid access to high-performance GPU clusters for training and inference, trusted by the world's leading AI labs and enterprises. Fluidstack was founded in 2017. The company is led by Gary Wu. Based in New York City, United States. Team size: 100-150. Total funding raised: $715.2M. Latest round: Series B ($450M, Jan 2026). Key investors include ["Cacti","Seedcamp","Mercuri","7 Global Capital","Episode 1 Ventures","7GC & Co","Google (in talks)","Situational Awareness (in talks)"].
- Founded
- 2017
- Headquarters
- New York City, United States
- Team size
- 100-150
- Total funding
- $715.2M
Value proposition
Enables deployment of multi-thousand GPU clusters in days rather than months with fully managed, single-tenant infrastructure purpose-built for AI workloads—eliminating setup complexity and delivering 95%+ of theoretical GPU performance.
Products and solutions
["Dedicated GPU Clusters (H100, H200, B200, GB200 with InfiniBand)","Atlas OS (bare-metal operating system for cluster management)","Lighthouse (real-time monitoring and predictive optimization platform)","On-Demand GPU Instances","Custom AI Data Center Solutions"]
Unique value
Distributed AI cloud model that aggregates GPU resources from independent data centers globally rather than owning hardware, enabling rapid scaling and competitive pricing. Deploys multi-thousand GPU clusters in 48 hours vs. 3-month wait times at traditional hyperscalers. Purpose-built exclusively for AI workloads with fully managed, single-tenant clusters.
Target customer
Frontier AI labs, AI research companies, enterprises developing large language models, and governments requiring high-performance AI infrastructure
Industries served
["Artificial Intelligence & Machine Learning","Large Language Model Development","Financial Services & Banking","Enterprise Software","AI Research & Development","Cloud Computing Infrastructure"]
Technology advantage
Proprietary technology stack combining Atlas OS (bare-metal OS automating server imaging, Kubernetes/Slurm provisioning, and cloud-like scaling) with Lighthouse (real-time monitoring, predictive optimization, auto-healing, and 15-minute engineering response SLAs). Delivers exascale compute with high-performance InfiniBand networking, achieving 95%+ of theoretical GPU performance. Single-tenant architecture ensures full hardware, network, and storage isolation.
How they differentiate
Distributed AI cloud model aggregating GPU resources from independent data centers globally rather than owning hardware, enabling deployment of multi-thousand GPU clusters in 48 hours vs 3-month wait times at traditional hyperscalers. Achieves 95%+ theoretical GPU performance with fully managed, single-tenant infrastructure purpose-built exclusively for AI workloads.
Main competitors
["CoreWeave","Lambda Labs","RunPod"]
Key partnerships
["Anthropic ($50B AI infrastructure partnership for US data centers)","NVIDIA, Dell Technologies, Borealis Data Center (exascale GPU cluster deployment)","VAST Data (AI data platform for global workloads)","DDN (high-performance storage solutions)","Google (TPU provider, reportedly considering $100M investment)","Hut 8, TeraWulf (data center infrastructure)","Macquarie Group (GPU financing)","Mistral AI, Character.AI, Poolside, Black Forest Labs, Meta (major customers)","Brookfield, BPIFrance, MGX (€10B France AI supercomputer project)"]
Notable customers
["Anthropic","Meta","Mistral AI","Character.AI","Poolside","Black Forest Labs","Midjourney","Samsung","Nvidia","Dell","Honeywell"]
Major milestones
["$50B Anthropic partnership for US data centers (Nov 2025)","€10B commitment for 1GW AI supercomputer in France (2025)","Europe's largest AI supercomputer deployment for Mistral AI (Mar 2025)","$6.7B contracted revenue through TeraWulf hosting agreements (Feb 2025)","$200M Series A led by Cacti (Feb 2025)","Gold tier rating from SemiAnalysis ClusterMAX™ 2.0 (Nov 2025)","Headquarters relocation to New York City (2025)","$700M funding discussions at $7B valuation (Dec 2025)","Google considering $100M investment (Feb 2026)"]
Growth metrics
Revenue grew from $1.8M (2022) to $30M (2023) to $66.2M (2024). Manages 100,000+ GPUs globally. Private cloud segment accounts for 62% of revenue with contract values averaging $100M+.
Market positioning
European AI champion and Gold-tier neocloud provider (per SemiAnalysis ClusterMAX™ 2.0 rating) competing against US-based providers. Focuses on frontier AI labs requiring exascale compute capacity with rapid deployment capabilities. Serves both private cloud (62% of revenue with $100M+ average contract values) and marketplace segments.
Geographic focus
Europe (UK, France, Iceland) and North America (US). Relocated global headquarters from London to New York City in 2025. Deploying data centers in Texas, New York, Iceland, and France. Strategic focus on European digital sovereignty while expanding US footprint.
Patents and IP
No registered patents publicly disclosed as of February 2026; competitive advantage derived from proprietary Atlas OS and Lighthouse platform, operational expertise, and strategic partnerships
About Gary Wu
Gary Wu is the CEO and Co-Founder of Fluidstack, and also Co-Founder of Treecard. He read Economics and Management at the University of Oxford and is a self-taught software engineer. He previously co-founded ventures with collaborators including Fluidstack, the AI cloud platform.
Official website: https://www.fluidstack.io