
Big swing, same playbook
Nebius just made its boldest geographic move yet: a £1.7 billion push to expand AI cloud capacity across four UK sites. Think of it like the company dropping anchor in one of Europe’s most important AI markets instead of just dabbling on the sidelines.
The buildout matters because this isn’t random capex cosplay. Nebius is expanding where enterprise demand, public-sector AI interest, and government support are all pulling in the same direction. In plain English: it’s building where the customers are likely to show up.
Nvidia is the secret sauce
Every site will run on Nvidia’s Blackwell Ultra infrastructure, which gives Nebius a pretty clear pitch to customers — not just raw GPU access, but a full-stack AI factory with proprietary software layered on top. That’s the kind of setup that makes hyperscalers look a little less automatic and a lot more expensive.
And this is where the story gets spicy for investors. Nebius already has heavyweight validation from contracts with Microsoft and Meta, plus customer use cases like Revolut running production workloads for AI training and fraud detection. So when the company says it wants more capacity, it’s not throwing darts at a wall. It’s trying to keep up with demand it can actually point to.
Why you should care
Nebius is also scaling fast enough to make your calculator sweat:
- Q1 2026 revenue jumped 684% year over year to $399 million
- Core AI business adjusted EBITDA margin expanded to 45% from 24%
- 2026 capex guidance was raised to $20 billion to $25 billion
That’s a company spending like a hyperscaler and growing like a startup, which is either thrilling or terrifying depending on your risk appetite. The UK expansion reinforces the idea that capital is following confirmed demand, not just hype.
Big picture: Nebius is trying to turn itself into the AI infrastructure equivalent of a fast-rising toll road — expensive to build, but potentially very valuable if traffic keeps piling in.
