NVIDIA deepens UK AI commitment with £2 billion startup fund and 65 MW infrastructure push at London Tech Week

At London Tech Week on June 7, NVIDIA founder Jensen Huang and UK Prime Minister Keir Starmer declared that "the U.K. would be an AI maker, not an AI taker" — a commitment backed by a cluster of infrastructure announcements, startup investments, and sovereign compute deployments that position the UK as a focal point of NVIDIA's European AI strategy.

What's new

NVIDIA's UK activity spans infrastructure, enterprise, and startup investment, all announced simultaneously at London Tech Week:

New compute deployments: AI cloud provider Nebius announced three new NVIDIA-powered data center deployments in the UK, reaching "65 megawatts when fully ramped up in 2027." BT and Nscale added a joint sovereign data center combining NVIDIA AI infrastructure, Nscale's full software stack, and BT's nationwide connectivity network.

Isambard-AI: Britain's national AI supercomputer is "built on 5,400 NVIDIA GH200 Grace Hopper Superchips and running entirely on zero-carbon electricity." Hosted at the University of Bristol, Isambard-AI serves researchers across more than 30 UK universities through NVIDIA's Deep Learning Institute program and has become a proving ground for inference optimization by UK AI startups.

£2 billion startup investment: NVIDIA committed a "£2 billion investment in the U.K. startup ecosystem" through its Sovereign AI Fund, with four NVIDIA Inception portfolio companies spotlighted:

• Cosine is building a sovereign coding platform for regulated industries, training a new mixture-of-experts multimodal agentic LLM
• Doubleword achieved "70x faster model cold starts" and "4x lossless KV cache compression" on Isambard infrastructure, and reports "inference at 90–95% lower costs than other leading inference providers"
• Cursive and Prima Mente round out the funded cohort

Developer ecosystem: More than 200,000 UK developers are now enrolled in NVIDIA's Developer Program.

Context

NVIDIA has been accelerating its sovereign AI strategy across multiple geographies in parallel. Similar partnership and infrastructure announcements came in June for South Korea and earlier in 2026 for Taiwan. The pattern is consistent: NVIDIA anchors national AI programs to GH200 hardware, funds local AI startups through the Inception and Sovereign AI Fund programs, and trains local researchers through the Deep Learning Institute.

The UK's AI Action Plan, published in January 2026, set explicit targets for making Britain a global AI hub. London Tech Week has become the venue where those policy commitments get matched with concrete vendor investments. Jensen Huang's direct appearance alongside PM Starmer elevates these announcements above typical partner press releases — the "AI maker, not an AI taker" framing is a deliberate positioning of the UK against European peers that have defaulted to licensing US-developed cloud services.

Isambard-AI itself reached operational status in late 2025 and is housed in a zero-carbon facility at the National Composites Centre in Bristol. It represents the largest single investment in publicly funded UK research compute in a generation.

Why it matters

The sovereign AI strategy is a structural play for NVIDIA. By embedding GH200 infrastructure into government-backed national compute programs, NVIDIA creates decade-long procurement relationships that are difficult to displace. When a national supercomputer runs on GH200s and local researchers train on it, the path of least resistance for those researchers — and the startups they spin out — is to stay on NVIDIA hardware.

For UK AI startups, the more immediate implication is access to frontier compute with favorable economics. Doubleword's "inference at 90–95% lower costs than other leading inference providers" figure, if it holds up at scale, is a real competitive advantage over teams building on US hyperscaler APIs. The combination of national compute access, Sovereign AI Fund backing, and DLI training creates a vertically integrated support structure for UK AI companies that few other markets can currently match.

With 65 MW of new capacity coming online by 2027 — combined with the existing Isambard-AI deployment — the UK is building meaningful AI infrastructure headroom at a time when global compute demand continues to outpace supply. That positioning matters for attracting AI investment and talent to the UK over the next five years.