About Anyscale
Anyscale was founded in 2019 by Robert Nishihara, Philipp Moritz, and Ion Stoica, researchers from the UC Berkeley RISELab who had created Ray as an open-source project to solve a specific problem in distributed computing: making it much easier for Python developers, particularly those working on machine learning, to scale their code from a single laptop to large clusters of machines without having to become experts in distributed systems engineering. Ray became widely adopted well beyond Anyscale itself, used inside major AI labs and companies to handle demanding workloads such as large-scale model training, hyperparameter tuning, and reinforcement learning, and it has been a foundational piece of infrastructure behind some of the largest language model training efforts in the industry.
Anyscale's commercial product, built on top of the open-source Ray project, provides a managed platform that handles the operational complexity of running Ray clusters in production, including auto-scaling, cluster management, cost optimization, and integration with major cloud providers, letting companies get the benefits of Ray without needing a dedicated infrastructure team to operate it themselves. As generative AI training and inference workloads have grown dramatically in scale and cost, Anyscale has positioned Ray and its commercial platform as core infrastructure for AI labs and enterprises building and serving large models, competing for this workload with cloud providers' own native distributed computing offerings as well as other open-source distributed computing frameworks. Ion Stoica, one of Anyscale's co-founders, is also a co-founder of Databricks, and Anyscale has maintained a close relationship with the broader open-source data and AI infrastructure ecosystem. The company has raised substantial venture funding from investors including Addition, Ostrom, and Foundation Capital, reaching a valuation exceeding a billion dollars as demand for scalable AI infrastructure has surged.
