SWI Group launches SWI Compute to build NVIDIA AI factories

A new infrastructure provider

SWI Group, a transatlantic investment firm listed on Euronext Amsterdam, has officially launched a new subsidiary called SWI Compute. This new entity is designed specifically to build and operate AI factories that utilize NVIDIA hardware. The company intends to focus its initial efforts on establishing these facilities across Europe.

The launch comes as the demand for high-performance computing continues to outpace what is currently available on the market. SWI Compute aims to address this shortage by providing dedicated infrastructure for the growing AI ecosystem. To support this, the company became a Preferred NVIDIA Cloud Partner in August.

SWI Group has been preparing for this expansion since 2020 by investing in land, power, and shell structures. The group currently holds 3.6 GW of power capacity across 24 different sites in both the United States and Europe. This foundation allows them to move quickly into the specialized AI market.

The UK factory and expansion

The company's first large-scale AI factory will be located in the United Kingdom. This site is scheduled to come online by the end of 2027 and will initially start with a capacity of 30 MW. However, the long-term plan for the UK campus involves a significant expansion to 530 MW.

This UK project serves as the first phase of a much larger roadmap for the company. SWI Compute plans to reach a total of 2.3 GW of NVIDIA AI factory capacity across Europe by 2029. The company will also utilize five existing European campuses located in Ireland, Denmark, Spain, and Italy.

The growth strategy is supported by a massive hardware commitment from NVIDIA. SWI Compute has secured approximately US$1 billion in advanced hardware orders to populate these sites. These orders include nearly 7,000 Rubin GPUs and 3,500 Vera CPUs for the new infrastructure.

Technical architecture and design

Every new site built by SWI Compute from 2027 onward will follow a specific blueprint. These facilities will be designed to NVIDIA reference architectures, specifically the Vera Rubin DSX AI Factory reference design. This means the hardware and software will be tightly integrated to optimize performance.

The company will use the full-stack NVIDIA DSX AI Factory Platform to design, build, and operate these sites. This approach provides a fully integrated infrastructure rather than just offering raw computing power. By following these reference designs, the company ensures the hardware is productive and durable for its customers.

Nico Caprez, Vice President of Global AI Infrastructure Growth at NVIDIA, noted that SWI Compute is utilizing European power sources to support the AI industry. He described the infrastructure being built as productive, durable, and fungible. This suggests the factories are designed to be flexible enough to handle various types of AI workloads.

Market trends and investment

The launch occurs during a period of intense investment in the European data center market. According to Arizton Advisory & Intelligence, the market saw investments of US$59.84 billion in 2025. This figure is projected to rise significantly to US$148.82 billion by 2031.

The United Kingdom is currently a primary destination for this type of capital. In 2025, the UK received the highest percentage of data center funding at 27 per cent. Germany followed as the second-largest recipient with 12 per cent of the total investment.

SWI Group is positioning itself as a vertically integrated digital infrastructure company through this move. By owning the land, power, and hardware, they can offer a more cohesive service to their first customers. This vertical integration is a response to the fact that the market continues to surpass available capacity.

What comes next for SWI

SWI Compute has already finalized contracts with its first customers to begin utilizing the new infrastructure. These early partnerships will serve as a test case for the 2027 UK launch. The company will focus on scaling the UK site from its initial 30 MW to the planned 530 MW capacity.

The broader goal remains the completion of the 2.3 GW capacity target by 2029. This will require the company to rapidly expand its footprint to meet the high demand for AI compute. They will continue to build out their five European campuses to support this growth.

As the company moves forward, it will be a key player in the transition toward specialized AI factories. By aligning with NVIDIA's specific reference designs, SWI Compute is betting on a standardized, high-performance model for the future of AI infrastructure. This strategy aims to provide a stable foundation for the next generation of AI models.