Lambda (formerly Lambda Labs) is a specialized AI cloud provider that rents GPU compute — primarily NVIDIA hardware — to AI researchers, enterprises, and hyperscalers. Customers get on-demand and reserved access to high-density GPU clusters housed in Lambda-owned and operated data centers, purpose-built for training large models and running inference. Lambda differentiates from general-purpose clouds (AWS, GCP, Azure) by focusing exclusively on AI workloads: lower pricing, simpler provisioning, and frontier hardware access. Its customer mix spans individual researchers, frontier AI labs, and large enterprises, including Microsoft under a multi-billion-dollar capacity agreement and NVIDIA itself — which leased back 18,000 GPUs from Lambda in a $1.5B arrangement. The company is building out what it calls "AI factories" — high-density, liquid-cooled data centers co-engineered for NVIDIA's latest Blackwell and Vera Rubin silicon. Data centers are now live or in development across Chicago, Los Angeles, Seattle, Kansas City, Quebec City, and Querétaro, Mexico. Lambda is targeting an IPO in the second half of 2026.