Fuse Energy launches UK AI compute service with NVIDIA GPUs

Bridging the power gap

Fuse Energy has officially opened reservations for Fuse Compute, a new service designed to provide access to NVIDIA GPUs within UK data centres. This launch signals a significant shift for the energy supplier as it moves into the commercial AI infrastructure market. The company aims to address a major hurdle in Britain: the difficulty of securing sufficient power for new data centre developments.

Demand for AI compute has surged recently, but new projects often face long delays when trying to secure connections to the national grid. Fuse Energy notes that these waits can stretch to a full decade in some instances. By combining its existing energy operations with data centre access, the company hopes to streamline the process of bringing new computing capacity online.

The UK Government has estimated that the country will require at least 6GW of AI-capable data centre capacity by the year 2030. Currently, domestic supply remains limited, which often forces UK-based AI firms to rely on infrastructure hosted in other countries. Fuse Compute intends to provide a domestic alternative that keeps data and legal control within the UK.

Technical rollout and scale

The initial rollout of Fuse Compute will feature 128 NVIDIA B300 GPUs, which are expected to be available starting in December. Following this first phase, the company plans to expand its capacity to 1,024 GPUs during the second quarter of next year. This growth is part of a larger roadmap that includes adding 22MW of capacity in the fourth quarter of next year and another 116MW starting from 2028.

The service is designed to provide bare-metal, single-tenant NVIDIA GPU nodes. This means customers get dedicated hardware rather than shared, off-the-shelf cloud access, which is often preferred by those with specific performance or privacy needs. Because the service operates under UK jurisdiction and uses a UK data path, it is positioned for customers who have strict requirements regarding where their data is stored and managed.

Pricing, delivery terms, and specific configurations will be negotiated on a per-contract basis. This approach suggests that Fuse Compute is initially targeting larger-scale customers or research groups that require dedicated capacity. By offering high-demand NVIDIA processors, Fuse is entering a market where hardware availability has become increasingly tight due to the global surge in AI model training.

Infrastructure and investment

Fuse Energy is leveraging its status as a major utility provider to build out this infrastructure more efficiently. The company provides electricity to 470,000 UK households and maintains both meter operator and supplier licenses. They also manage a power generation pipeline of more than 1GW across the UK, the US, and Europe, alongside a data centre connection pipeline of 142MW across 15 different projects.

The company claims that its control over energy assets and grid connections can significantly shorten the time required to launch AI compute. As evidence of its efficiency, Fuse noted that its most recent solar site was completed at a price roughly 30% lower than the standard industry cost per megawatt. This suggests they can deliver the necessary energy infrastructure at a lower cost than some of their competitors in the space.

The expansion into AI infrastructure involves a significant financial commitment to the UK economy. Fuse reports that its near-term data centre projects represent GBP £1 billion of investment in UK infrastructure. These projects are also expected to create 250 construction and operational jobs, further diversifying the company's portfolio beyond its core consumer and energy operations.

Sovereignty and local growth

The move into domestic compute is driven by growing concerns over the UK's dependence on overseas infrastructure. Policy makers and technology companies have expressed worries about the resilience and sovereignty of AI development when it happens outside of the country. Limited home-grown supply makes it difficult for British AI firms to scale their operations locally without relying on foreign jurisdictions.

By operating as both a power provider and a compute host, Fuse is entering a market where infrastructure constraints are just as important as the chips themselves. Access to land, electricity, and grid connectivity are now the primary factors determining which projects can move forward quickly. Fuse aims to simplify these complexities for businesses that need to build the future of intelligence within British borders.

Alan Chang, the Chief Executive Officer and Co-Founder of Fuse Energy, stated that "the grid can't connect data centres fast enough to keep up." He noted that Fuse Compute puts their expertise to work for companies building AI. The goal is to ensure that the development of new intelligence technologies happens domestically rather than being outsourced due to power shortages.