EP26 - Nscale: The European Hyperscaler for AI

From Series A of $155 million to Series B of $1.1 billion, the Microsoft deal for 200,000 GPUs and the $1 billion joint venture with Aker and OpenAI in Norway: Nscale's full-stack playbook, the European hyperscaler building the continent's sovereign AI infrastructure.

Nscale is not a generalist cloud provider — it's vertical infrastructure built specifically for AI workloads: high-density GPU clusters, networking optimized for training and inference, and proprietary software that maximizes resource utilization. The thesis is that AWS, Azure, and GCP have architecture optimized for traditional workloads and are under-optimized for AI, opening a window of opportunity for a specialized player.

In this episode we analyze the cost advantage of the vertical model datacenter + GPU + software, the industrial partnership strategy that drove the mega-deal with Microsoft and the JV in Norway (where renewable energy is abundant and data stays in Europe), and Serverless Inference as the wedge to acquire AI developers. We discuss why data sovereignty has become a real sales argument for European companies, and how Nscale is positioning infrastructure as both technical and political response to dependence on American hyperscalers.

• Verticale DC+GPU+software come vantaggio di costo
• Partnership industriali per sbloccare megadeal
• Serverless Inference come wedge per i developer