ANNOUNCEMENT: SWI Compute launches with ~ $1 Billion of NVIDIA systemsLaunching with ~$1B of NVIDIA systems

NVIDIA AI factories · Starting in Europe

The foundation is built.Now comes compute.

SWI Compute turns gigawatts of power into NVIDIA AI factories, with dedicated and on-demand capacity for Europe’s AI natives, AI labs and enterprises.

Read the announcement

3.6GW

Power platform across Europe and the US

~$1B

NVIDIA systems ordered in our first 60 days

European campuses: Ireland, UK, Denmark, Spain, Italy

Preferred

NVIDIA Cloud Partner

Intelligence at homeWho it’s for

Built for the companies that need capacity most

Europe has the researchers, founders and enterprises to build frontier AI. The capacity that exists has gone first to the buyers able to sign multi-year contracts at scale. AI natives with frontier ambitions have the hardest time securing it. SWI Compute is built for them.

AI natives

Frontier ambitions without a hyperscaler’s balance sheet. On-demand access when you need it, dedicated capacity when you scale.

AI labs

Train, fine-tune and serve on the same NVIDIA infrastructure, close to your users and your data.

Enterprises

AI in production in Europe, under European jurisdiction.

Six years ahead on powerSince 2020

Power is the biggest constraint on AI in Europe. We secured it first.

20202026Land, power, shellContracted, on order

SWI Group began investing in land, power and shell in 2020. Today the power is contracted and the systems are on order.

The group develops its own sites, from grid connection to commissioning.

Built with NVIDIA and its ecosystemReference architecture

Every site built to NVIDIA reference architecture

Every new site from 2027, starting with Cambridge, follows the NVIDIA Vera Rubin DSX AI Factory reference design, which codesigns compute, networking, power and cooling as one system.

First large-scale AI factory

Cambridge

The first phase of a campus planned to grow to 500 MW, and the first step toward 2.3 GW of NVIDIA AI factory capacity in Europe by 2029.

Capacity: 30 MW; First phase, growing to 500 MW
NVIDIA GPUs: ~7,000; Close to 7,000 Rubin GPUs
Systems: Vera Rubin NVL72 racks, DSX reference design
Online: By end of 2027

Cloud partner

Preferred partner in the NVIDIA Cloud Partner program

Systems partner

Delivering more than 8,000 GPUs

Systems integrator

Integration and deployment

Productive, durable, fungibleWhy it lasts

Capacity that stays current

Per megawatt

Productive

Codesign across the full stack delivers the highest throughput and the lowest cost per token.

Over time

Durable

Continuous software optimization keeps installed systems earning years after they ship.

Across workloads

Fungible

From pretraining to agentic inference, alongside simulation, data processing and scientific computing.

Power is the biggest constraint on AI in Europe. SWI Group spent six years securing it, and SWI Compute is putting it to work in NVIDIA AI factories.And we’re just getting started in Europe and beyond.

Halil MirakhmedovCEO, SWI Compute

NewsAll news

Press release · Amsterdam · Singapore · Oct 26, 2026

SWI Compute launches and turns gigawatts of power into NVIDIA AI factories, starting in Europe

~ $1 billion of NVIDIA Blackwell Ultra and Vera Rubin systems ordered within 60 days of becoming a Preferred NVIDIA Cloud Partner.

Read the release

Latest from SWI Group

12 Oct 2026 · SWI GroupSWI Group announces €800 million institutional capital increase at a €6.1 billion valuation 30 Sep 2026 · SWI GroupSWI Group reports H1 2026 results as it accelerates into AI infrastructure and compute 15 Jun 2026 · SWI GroupSWI Group takes majority position in GDA, a 1.3 GW US digital infrastructure group

Request capacity

Tell us what you are
building.

Dedicated or on-demand, our team will come back with what we can offer and when.

On-demand access: When you need it
Dedicated capacity: When you scale