UK-US Tech Prosperity Deal to deliver 120,000 GPUs and Europe's largest AI factory

UK-US Tech Prosperity Deal: Building Europe's largest AI compute hub

The UK and US have agreed the Tech Prosperity Deal to accelerate artificial intelligence, quantum and nuclear technologies. The plan concentrates on scale: 120,000 GPUs are set to be deployed in the UK within 12 months, backed by £44bn of UK government investment and £31bn committed by Microsoft, Nvidia, Google, OpenAI and CoreWeave.

Government's message is clear: attract founders, anchor high-growth startups, and make compute capacity abundant. The goal is for Britain to host Europe's largest AI "factory" through GPU clusters, with Labour also positioning the UK as Europe's largest gigafactory.

What's in the pact

• Joint UK-US research on AI for healthcare (targeted treatments for cancer and rare/chronic diseases) and shared priorities like fusion energy.
• Compute build-out: Nvidia to supply 120,000 GPUs to the UK over 12 months.
• OpenAI: first phase of 8,000 GPUs, scaling to 60,000 Nvidia Grace Blackwell Ultra GPUs.
• Microsoft: £22bn investment, including 23,000 advanced GPUs to deliver an AI supercomputer in Loughton.
• Google: £5bn for a new datacentre in Waltham Cross.
• Nscale involved across deployments, including an OpenAI Stargate project in Norway using 100,000 Nvidia chips.

Why this matters for government leaders

• Compute access: Departments, NHS bodies and agencies can plan pilots and production systems with far more AI capacity onshore. Expect new procurement routes and cross-government allocations.
• Mission outcomes: Opportunities in diagnostics, long-term condition management, fraud detection, border operations, and infrastructure planning.
• Regulation and assurance: Scale requires stronger model testing, audit trails, safety evaluations and incident reporting aligned to UK policy.
• Data stewardship: High-value public datasets will be in scope for secure training and fine-tuning. Strong privacy, IP and security controls are non-negotiable.
• Regional growth: Siting of datacentres and supercomputing will affect skills pipelines, energy and planning decisions across local areas.

Described as "the first-ever UK-US tech deal", Kanishka Narayan said the agreement "has the potential to transform lives right across Britain." He added the government will be "laser-sharp focused on the execution of the skills element," with a push to convince "the very best founders across the world that Britain is the right place for them to build."

Immediate actions for departments and agencies

• Map your top three AI use-cases that benefit from larger models or fine-tuning (e.g., triage, forecasting, casework summarisation) and define success metrics.
• Engage your CDO, CISO and commercial teams to prepare data access, privacy impact assessments and contract terms for GPU-backed services.
• Identify candidate datasets for secure training and evaluation. Establish red-teaming and model risk testing plans.
• Coordinate with DSIT, NHS England, MoD/Dstl or relevant leads to align with shared research programmes.
• Plan energy, cooling and resilience requirements for on-prem or hybrid deployments linked to new facilities.

Skills and capability

Government will lean on universities, researchers and startups, backed by a 50-point AI opportunities plan. Build internal capability while partnering with industry for specialist roles like AI safety, evals, MLOps and data engineering.

For team upskilling by job family, see curated AI learning paths and certifications that map to practical roles and tasks. Explore courses by job function.

Risks to manage

• Concentration risk: Dependence on a small set of vendors and architectures. Mitigate with multi-vendor strategies and portability.
• Safety and reliability: Evaluation, alignment, and monitoring must keep pace with larger models and higher stakes use-cases.
• Privacy and security: Strict controls for sensitive and citizen data, plus secure enclaves and auditability for training and inference.
• Energy and sustainability: Plan grid impact, water use and carbon reporting for datacentres and supercomputing sites.
• Workforce pipeline: Scale up apprenticeships, fast-track recruitment and cross-government communities of practice.

Deployment timeline and key sites

• Next 12 months: 120,000 GPUs delivered to the UK; compute capacity ramps in phases.
• Loughton: Microsoft's UK AI supercomputer with 23,000 advanced GPUs.
• Waltham Cross: Google's £5bn datacentre investment.
• Partnerships: OpenAI and Nscale deployments in the UK; Nscale's related build in Norway indicates broader supply coordination.

Policy context and further reading

• UK AI regulation: a pro-innovation approach (policy paper)
• UKRI - research funding and programmes

"We're laying the foundations for a future where together we are world leaders in the technology of tomorrow," said prime minister Keir Starmer, highlighting highly skilled jobs, local benefits and a nationwide focus on deployment. With compute scaling fast, the window for departments is clear: pick high-value use-cases, secure data pipelines, and move from pilots to delivery.