Nscale will absorb the Anyscale team across the United States, Europe, and India, while maintaining the brand as a distinct entity for existing customers. The acquisition centers on the Ray open-source framework, which Anyscale developers pioneered for distributed AI applications. By joining the PyTorch Foundation, Nscale intends to support the continued evolution of this standard, ensuring that the software remains accessible to the broader developer community even as it anchors a proprietary cloud offering.

For Nscale CEO Josh Payne, the move is a departure from the traditional infrastructure model where providers simply rent out GPU capacity. Instead, the company manages the entire stack, from physical data centers and power supply to the software that executes training and inference workloads. Anyscale’s platform provides the managed services necessary to bridge that infrastructure gap, allowing clients to fine-tune large language models and deploy agents using their own proprietary data.

The transaction, which involves undisclosed financial terms, remains subject to customary regulatory approvals and closing conditions. If finalized in the second half of 2026 as expected, the merger will allow Nscale to offer a unified environment for AI development, potentially reducing costs and complexity for enterprises moving from experimental AI to large-scale production.