European cloud operator Nebius has launched a new GPU cluster near London, UK.

The company this week announced its first deployment of Nvidia Blackwell Ultra AI infrastructure in the UK. The cluster sits in Ark Data Centres’ Longcross Park campus in Surrey.

The cluster totals 126 racks across three data center halls. The first phase comprises 4,000 Nvidia HGX B300 GPUs, housed within Nebius’ self-developed Gen5 servers. The second phase will comprise another 3,000 B300s.

The deployment totals 16MW of capacity. The air-cooled, hot-aisle contained clusters feature Nvidia Quantum-X800 InfiniBand network connectivity. The cluster was deployed in just two months.

News of the deployment was first announced in June, with Ark naming itself as the host shortly afterwards.

In a statement, Arkady Volozh, founder and CEO of Nebius, said: “This launch is a milestone for Nebius and for the UK’s growing AI ecosystem. By combining our cloud infrastructure foundations with Nvidia’s latest technology, we’re empowering organizations across the UK to train, deploy, and scale AI models and applications faster, more securely, and more sustainably than ever.”

At the event, Volozh added that the UK is one of the “global centers of AI,” which is one of the reasons the company chose to deploy here. He also noted that amongst regular announcements of huge amounts of promised capacity, Nebius’ deployment was “real capacity, here today.”

Customers of Nebius set to use the cluster include life sciences AI startups Basecamp Research, Prime Mente, and Sword Health.

In a statement, Basecamp Research co-founder, Oliver Vince, said: “Tackling unprecedented challenges like designing programmable medicines requires infrastructure that can handle massive proprietary datasets from across the globe while maintaining the speed and reliability to push scientific boundaries.

“Nebius’s infrastructure delivers the scale and performance we need; having it here in the UK means we keep greater control over our data and can compete at the cutting edge of AI-driven life sciences.”

At the cluster launch, Hannah Madan, co-founder at Prima Mente, said that for companies such as hers developing their own AI models, data centers act as “science discovery centers.”

“It would not have been possible to build this company five years ago,” she said.

Nebius was formed last year after Russian tech firm Yandex’s European operations were spun off into a separate entity. Based in Amsterdam, Nebius took control of Yandex’s Finnish data center, its Nebius AI unit, as well as data firm Toloka AI, edtech provider TripleTen, and autonomous driving firm Avride.

The company is aiming to achieve a baseline of 100MW of capacity in 2025, but according to COO Roman Chernin, the ambition is actually to grow “much larger and much faster,” with the company looking to build a data center pipeline to provide “scalability to more than 1GW of power.”

In the US, the company is working with DataOne to develop a 300MW data center in New Jersey, the first phase of which is expected to go live in the summer of 2025. Nebius is also set to lease space at an upcoming data center in Kansas City, Missouri.

The company has further colocation presence in Iceland and France, the latter in an Equinix facility in Paris. In May 2025, Nebius revealed it would build and operate a $140 million national supercomputer in Israel, part-funded by the Israeli government’s Innovation Authority.