Production-scale AI with Ray — the commercial platform for distributed Python & LLM workloads
HQ: San Francisco, CA, United States | Founded: 2019 | Employees: 200–430 (est.; 430 per Latka/Tracxn 2023) | Stage: Series C+ ($1B valuation) → Acquired by Nscale ($1.65B, 2026) | Website: https://www.anyscale.com
Anyscale was founded in 2019 in San Francisco by Robert Nishihara (CEO), Philipp Moritz, and Ion Stoica — creators of Ray at UC Berkeley's RISELab (AMPLab successor) — along with Berkeley professor Ion Stoica (also co-founder of Databricks/Conviva). Ray is the open-source distributed computing framework that lets AI builders scale Python workloads (training, batch inference, RL) across clusters without rewriting for Kubernetes. Anyscale is its commercial platform: managed Ray on any cloud with autoscaling, observability and LLM serving. It raised $20.6M Series A led by a16z (2019 press), $40M Series B led by NEA (Oct 21 2020), $100M Series C at $1B (Dec 7 2021, Venture Capital Journal/WSJ), plus $99M addition in Aug 2022 from Addition/Intel per WSJ at >$1B valuation. Tracxn records $281M total; Latka pegs $111.9M ARR in 2023 on 430 employees. In Jul 2024 Robert Nishihara stepped to co-founder/product as Keerti Melkote became CEO; in 2025-26 Bloomberg/JFiance reported Anyscale agreed to be acquired by Nscale (UK AI compute infra) for $1.65B.
Anyscale is the AWS of Python scale: Ray made 'for i in range(1000000): ray.get(fn.remote())' run across any cloud, and Anyscale is the managed commercial skin. Founded 2019 by RISELab PhDs Robert Nishihara, Philipp Moritz and Ion Stoica (the Ray authors) with Andreessen Horowitz leading $20.6M A to democratize distributed programming, it added $40M B led by NEA (Oct 2020), then $100M C at $1B (Dec 2021) plus $99M extension (Aug 2022, Addition/Intel) totaling $281M, with $111.9M ARR in 2023 (Latka). In 2024 founder-CEO Robert Nishihara ceded CEO to Keerti Melkote; in 2025-26 Anyscale agreed to be acquired by Nscale for $1.65B (JFiance/LI). The moat is Ray's open-source gravity (100M+ downloads) — the risk is hyperscalers (AWS SageMaker, Databricks) making distribution the product.
Browse: Home | Blog | About | All Startups A-Z | View full profile on StartupWiki