SWI Compute emerges with 3.6GW AI factories, $1B Nvidia order

SWI Compute has emerged from stealth to build and operate Nvidia-powered AI factories, starting in Europe. The company, a spin-out of Euronext-listed SWI Group, brings 3.6 gigawatts of power capacity secured over six years across Europe and the US, positioning itself as a solution to the continent’s AI infrastructure shortage.

The company became a Preferred NVIDIA Cloud Partner in August and has already placed orders worth approximately $1 billion for Nvidia systems. By 2029, SWI Compute aims to deploy 2.3 gigawatts of Nvidia AI factory capacity globally.

SWI Group’s head start in securing power gives SWI Compute a rare advantage. While hyperscalers and well-funded AI labs dominate existing capacity, Europe’s AI startups and research institutions struggle to access compute resources locally. SWI Compute is targeting this gap, offering dedicated and on-demand access to Nvidia AI factories under European jurisdiction. Early customer interest suggests demand for its model, which prioritizes proximity to users and data while complying with local regulations.

The infrastructure follows Nvidia’s reference architectures, with future sites built on the Nvidia DSX AI Factory Platform. The UK campus will use Nvidia systems connected via Nvidia Quantum-X800 InfiniBand and Spectrum-X Ethernet networking. Delta Computers is serving as the systems integrator, alongside storage partners, to deploy thousands of GPUs in the initial phase. Nvidia’s Nico Caprez called the approach "productive, durable and fungible," emphasizing the long-term value of infrastructure that can adapt to new workloads and customers.

Europe’s AI ecosystem has long relied on capacity outside the region, a dynamic SWI Compute aims to change. The company’s pitch centers on reducing latency and regulatory friction. Halil Mirakhmedov, CEO of SWI Compute, framed energy availability as the primary constraint for European AI infrastructure, noting that demand continues to outpace supply.

The timing aligns with Nvidia’s growing focus on AI factories, a concept Jensen Huang has championed as a shift from silicon to deployment challenges. When StartupReader covered Nvidia’s $8 billion European investments last month, the emphasis was on backing startups like Mistral AI and Nscale. SWI Compute’s emergence suggests Nvidia is also deepening its infrastructure partnerships to support those startups at scale.

SWI Compute’s model hinges on converting power into long-term revenue. AI factories generate value through efficiency, workload versatility, and software optimization that extends hardware lifecycles. For customers, this means access to current capacity; for SWI Group, it transforms energy assets into a cloud platform.

The next phase begins in late 2027, when additional AI campuses are scheduled to come online. With 3.6 gigawatts of power already secured, SWI Compute’s challenge will be execution—scaling deployments while maintaining the efficiency gains promised by Nvidia’s integrated systems. If successful, it could redefine how Europe’s AI startups access compute, shifting the balance from remote hyperscalers to local, purpose-built infrastructure.

What to watch next is whether SWI Compute can attract AI startups and labs at scale. The company’s power advantage is clear, but its ability to serve smaller players—not just hyperscalers—will determine its role in Europe’s AI ecosystem. The first UK campus, expected online by the end of 2027, will be an early test.

Sources: finance.yahoo.com