Everyone’s talking about the latest AI models. The chatbots. The agents. The demos that make your jaw drop. But here’s the thing nobody at your dinner party is discussing: none of it works without an absurd amount of GPU compute, and a scrappy UK startup just raised the largest single equity round in European history to make sure that compute exists.

On Monday, Nscale announced a $2 billion Series C at a $14.6 billion valuation — more than doubling its Series B from just six months ago. The round was led by Norway’s Aker ASA and 8090 Industries, with participation from Nvidia, Dell, Citadel, Jane Street, Lenovo, Nokia, and Point72.

If that investor list reads like a who’s who of “people who understand where the real money in AI is,” that’s because it is.

What Even Is a Neocloud?

If you haven’t encountered the term yet, buckle up — it’s about to be everywhere.

Neoclouds are purpose-built GPU cloud providers. Unlike AWS, Azure, or Google Cloud, they don’t try to be everything to everyone. They’re laser-focused on one thing: getting as many high-end GPUs as possible into racks and making them available to AI companies that are desperately hungry for compute.

The neocloud market sits at roughly $35 billion today and is projected to approach $240 billion by 2031. That’s not a niche. That’s an industry.

Nscale sits alongside competitors like CoreWeave (which IPO’d earlier this year), Crusoe, and Nebius (the Yandex spinoff that just signed a $3 billion deal with Meta). What sets Nscale apart is its “ground to cloud” approach — the company owns and operates its own data centers, GPUs, networking, and software stack. Full vertical integration that most competitors lack.

From Crypto Mining to AI Kingpin

Here’s the part that always gets buried: Nscale spun out of Arkon Energy, a crypto mining infrastructure company, in 2024. Yes, the crypto-to-AI pipeline is real, and it’s arguably one of the most successful pivots in recent tech history.

The logic is straightforward. Crypto mining operations already had the three things AI compute needs: power capacity, cooling infrastructure, and GPU supplier relationships. When AI made GPU compute dramatically more valuable than mining, the smart operators pivoted fast.

Founded by Josh Payne, Nscale has been on a funding tear that defies belief:

  • September 2025: $1.1B Series B (Nvidia-backed)
  • October 2025: $433M pre-Series C SAFE round
  • October 2025: $14B expanded partnership with Microsoft
  • February 2026: $1.4B delayed draw term loan
  • March 2026: $2B Series C at $14.6B valuation

That’s roughly $19 billion in committed capital in 18 months. For a company that’s barely two years old.

Why Microsoft Needs Outside Help

Here’s the counterintuitive part: Microsoft — one of the world’s biggest cloud providers — is one of Nscale’s biggest customers. Microsoft has signed more than $33 billion in capacity agreements with neoclouds collectively. A company that literally builds data centers for a living still needs outside help.

Why? Because AI demand is growing faster than even the hyperscalers can build. Training a frontier model requires tens of thousands of GPUs running for months. Inference requires even more aggregate compute as AI products scale. The hyperscalers can’t build fast enough, so they’re outsourcing overflow to neoclouds.

Nscale specifically handles AI inference in Europe for Microsoft — a signal that data sovereignty is becoming a critical factor as AI regulation tightens globally.

The Stargate Connection

Nscale partnered with OpenAI to launch a Stargate-branded AI data center in Norway, deploying 100,000 Nvidia GPUs. This feeds into the broader Stargate initiative — the Trump administration-backed plan to invest up to $500 billion in AI infrastructure across the US and allied nations.

The company currently operates across six countries: the UK, US, Norway, Portugal, Iceland, and is expanding into Asia. An $865 million investment in Madison, North Carolina adds hundreds of megawatts of new capacity to its North American footprint.

The IPO Is Coming

You don’t add Sheryl Sandberg, Nick Clegg, and Susan Decker to your board for the vibes. Nscale has already hired Goldman Sachs and JPMorgan as underwriters. The timeline hasn’t been set, but given the trajectory, a 2026 IPO looks likely.

CoreWeave’s IPO earlier this year gave the market a template for neocloud valuations. If Nscale can demonstrate the revenue growth that its $14 billion Microsoft partnership implies, this could be one of the biggest European tech IPOs in years.

The risk? Neoclouds are brutally capital-intensive. That $2 billion goes right back into facilities and systems. These companies are in an arms race to deploy GPU capacity, and winners are determined by who secures power, land, and chips fastest — not who has the best pitch deck.

Why This Matters

Nscale’s raise crystallizes five trends worth watching:

Infrastructure is the real bottleneck. We have more AI models than compute to run them. Companies controlling GPU capacity hold enormous leverage.

The neocloud layer is permanent. AI demand grows faster than any single player can build, creating structural need for specialized GPU providers.

Europe is carving out AI infrastructure. While the US and China dominate model development, European companies are building the physical layer — boosted by data sovereignty rules and abundant renewable energy in places like Iceland and Norway.

The crypto-to-AI pipeline works. Nscale is the most dramatic example of crypto infrastructure companies successfully pivoting to AI compute.

Nvidia’s ecosystem flywheel spins on. By investing in neoclouds, Nvidia ensures GPU demand while creating companies structurally dependent on its hardware. Competitors are struggling to replicate it.

The Bottom Line

While everyone argues about which chatbot is smartest, companies like Nscale are quietly building the physical infrastructure all AI depends on. A two-year-old company doesn’t reach a $14.6 billion valuation without something real underneath — and what’s underneath is the simple truth that AI needs more compute than currently exists.

The question isn’t whether neoclouds will be important. It’s whether the current buildout pace can keep up with demand. If it can’t, the companies controlling GPU capacity will be the OPEC of the AI age.