🤖 AI Summary
U.K. cloud startup Nscale — spun out of crypto-mining firm Arkon and only publicly visible since 2023 — was named this week as an infrastructure partner by Nvidia, Microsoft and OpenAI, sealing a string of heavyweight deals that accelerate Europe’s AI buildout. Highlights include a five-year, $6.2B pact with Microsoft and Aker to build hyperscale AI infrastructure in Norway, Microsoft’s commitment to a Loughton supercomputer initially hosting 23,040 Nvidia Blackwell GPUs (50 MW, scalable to 90 MW) by Q1 2027, OpenAI’s Stargate U.K. first phase deploying 8,000 GPUs (expandable toward ~31,000), Nvidia’s broader £11B push with Nscale/CoreWeave, and Nvidia CEO Jensen Huang’s £500M equity investment into Nscale. Nscale has also pledged $1B to Norway and previously targeted 100K–350K GPUs across its pipeline.
Technically and strategically, Nscale’s rapid rise matters because it pairs high-density GPU fleets, power/real-estate capacity and proprietary software to offer end-to-end GPU-as-a-service—addressing Europe’s “sovereign AI” demand for onshore compute and the fragmented market for hyperscale projects. The deals validate Blackwell-class GPU deployments at multi-megawatt scale and underscore supply, capital and power as the gating factors for large LLM training/inference. Risks remain: enormous capex/debt needs (CoreWeave’s multi‑billion debt raises are a cautionary benchmark), GPU supply logistics and the operational challenge of hitting aggressive GW/MW and GPU-count targets on tight timelines.
Loading comments...
login to comment
loading comments...
no comments yet