Nscale

Nscale is a British artificial-intelligence cloud provider, founded and led by chief executive Josh Payne and headquartered in London, that builds and operates the data centers its customers train and run models on. It belongs to the category the industry calls a neocloud: vertically integrated infrastructure spanning GPU compute, networking, storage and orchestration software, rather than a general-purpose cloud that added machine learning later. Its self-serve stack includes on-demand GPU instances, managed Slurm and Kubernetes, fine-tuning, and inference endpoints.

The company raised $2 billion in a Series C round in March 2026, led by Aker ASA and 8090 Industries with participation from NVIDIA, Nokia, Dell, Lenovo, Citadel, Jane Street and Point72 — described by the company as the largest single equity raise in European history. That followed a Series B of $1.1 billion led by Aker in September 2025. As part of the Series C, the Aker Nscale joint venture rolled fully into the company, with Aker remaining a leading shareholder and its chief executive on the board.

Capacity is concentrated in the Nordics and expanding into the United States and United Kingdom, with company-run sites at Glomfjord and Narvik in Norway, Loughton in England, and Texas and West Virginia in the United States, alongside partner-run sites in Portugal, Iceland and North Carolina. Microsoft commissioned 1.35 gigawatts of capacity in March 2026, to be delivered on NVIDIA Vera Rubin systems, in a partnership reported at $14 billion. In August 2026 Anthropic agreed a six-year arrangement reported at $45 billion, drawing on the West Virginia site from late 2027.

In July 2026 Nscale acquired Anyscale, the company behind the open-source Ray framework, for a reported $1.65 billion, adding the scheduling layer machine-learning engineers actually work in to infrastructure it already owned. Ray was donated to the PyTorch Foundation in 2025 and remains community-governed, with Nscale joining the foundation.

Nscale is reported to have hired Goldman Sachs and JPMorgan for an initial public offering seeking $3 billion on a United States exchange. The company is a supplier of compute rather than a maker of AI models or applications — its customers are the labs and hyperscalers, and its business is the physical capacity underneath them.