On July 30, 2026, Nscale announced a definitive agreement to acquire Anyscale, the platform company built by the creators of Ray. Deal terms were not disclosed. Anyscale’s roughly 200 employees across the United States, Europe, and India are to join Nscale, and the transaction is expected to close in the second half of 2026.
Anyscale provides managed infrastructure for scaling AI workloads across thousands of GPUs, spanning data processing, training, inference, and reinforcement learning, with customers in healthcare, e-commerce, and robotics. Nscale operates data centres and GPU compute. Nscale CEO Josh Payne said “Anyscale extends that offering with managed services that AI teams use to scale any workload, completing a truly vertically integrated AI cloud platform.” Anyscale CEO Keerti Melkote said the combination creates “the first full-stack AI hyperscaler.”
The deal is a clear statement about where margin sits in AI infrastructure. Renting bare GPUs is a commodity business competing directly on price with the hyperscalers; owning the orchestration software that decides how those GPUs are used is stickier and harder to substitute. Ray is one of the most widely adopted open-source frameworks for distributed AI workloads, so acquiring its commercial steward buys Nscale a position in the software layer above its own metal rather than another block of capacity.