SWI Compute — the AI‑compute arm of Euronext‑listed SWI Group — emerged from stealth on October 26, 2026 with a plan to build and operate large NVIDIA‑reference “AI factories” across Europe. The company says it has secured 3.6 gigawatts of power capacity and ordered approximately $1 billion of NVIDIA systems after joining NVIDIA’s Cloud Partner programme. The first large site announced is a UK factory that will start at 30 MW with close to 7,000 Rubin GPUs and 3,500 Vera CPUs, and is expected to come online by the end of 2027; that 30 MW is the initial phase of a UK campus planned to expand to 530 MW as SWI targets 2.3 GW of capacity by 2029.

What SWI announced

According to the company press release and its newsroom, SWI Group has spent six years securing grid access and land, and the new SWI Compute business will design, build and operate AI campuses to NVIDIA DSX reference architectures. SWI says its initial customer roster already includes a US‑based AI inference platform and that systems and networking (NVIDIA Quantum‑X800 InfiniBand and Spectrum‑X Ethernet) are on order for the first phase. SWI framed the move as addressing a shortage of local, large‑scale GPU capacity for European AI developers.

What this means for the market

For the automotive sector — and especially software and AI teams in Central and Eastern Europe — the arrival of Europe‑based AI factories matters because model training, large‑scale simulation and inference for ADAS, software‑defined vehicles and autonomous systems require vast GPU capacity. Local, jurisdictionally European compute reduces the need to export sensitive training data outside the region and can shorten turnaround for large experiments and retraining cycles. SWI and NVIDIA emphasised the intention to serve European AI natives, labs, universities and enterprises with capacity that sits “close to their users and their data.”

Manufacturers, tier‑1 suppliers and start‑ups that develop perception models, simulation pipelines or vehicle software could gain faster access to large‑scale training and inference without contracting hyperscalers abroad. That matters in Central and Eastern Europe where automotive R&D and software hubs have grown sharply in recent years — but whether SWI’s campuses open sites in CEE countries, or how capacity will be allocated and priced for automotive workloads, are company decisions not yet published.

SWI’s CEO Halil Mirakhmedov said the firm’s multi‑year effort to secure power was aimed at overcoming Europe’s primary bottleneck for AI infrastructure; NVIDIA’s Nico Caprez described the deployments as “essential infrastructure” built to NVIDIA reference designs. Those quotes underline the infrastructure focus that could enable regional AI workstreams — from data‑heavy ADAS model training to large simulation farms for software‑defined vehicle testing.

Operational details, site‑by‑site timelines, and commercial terms matter for automakers and suppliers. SWI’s announcement names planned campuses in Ireland, the UK, Denmark, Spain and Italy as its current deepest European presence; the company says it will deploy globally but with strongest presence in Europe. Automotive groups and suppliers in CEE should watch whether SWI expands into Eastern Europe or partners with local data‑centre and industrial players.

B2B vehicle marketplace on WorldTrader24 →

Sources: PR Newswire, SWI Group / SWI Compute, DatacenterDynamics, Morningstar