With Donald, Jenson and Sam in London this week, many eyes will be cast in the direction of the UK’s AI future, and the US’s influence in that. Their visit was certainly one of the key talking points at DCD Connect London today.

Jenson humorously recently referred to the UK as “having one of the richest AI communities anywhere on the planet, yet being the largest AI ecosystem in the world without its own infrastructure,” and as usual, Jenson isn’t wrong. It is one of my major frustrations with the UK – we’ve great at ideas and vision, but we’re not great at execution and delivery.

When Matt Clifford unveiled the UK’s AI Opportunities Action Plan earlier this year, the ambition was bold and clear: transform Britain into an AI superpower, securing a share of the £400 billion global AI prize. Central to that vision was a commitment that the UK must become an AI maker, not an AI taker – a nation that builds and exports the technologies shaping our future, rather than one that imports and consumes them.

The ambition is sound. The challenge is once again, execution. From where I sit, leading a UK-headquartered data centre operator it is evident that unless we tackle systemic barriers around energy, copyright, and infrastructure funding, the UK risks falling behind in the global AI race. 

This is not an abstract risk. Decisions being made today by investors, hyperscalers and AI developers will define where the world’s most advanced AI is trained and deployed. At present, those decisions increasingly favour the United States, the Nordics and Saudi Arabia – not Donald, Jenson and Sam’s destination this week. 

Energy: The Achilles’ Heel of AI Growth

Every conversation about AI development must start with energy – it’s the first domino in the run. Training frontier models is one of the most energy-intensive processes in modern computing, yet the UK is one of the most expensive places in the world for power. Running 1GW of capacity costs £1.8 billion annually here. In the US, Nordics and 27 other European nations, it is a fraction of that figure.

The reasons are structural:

• Outdated pricing frameworks mean electricity bills remain still tied to volatile gas prices, even though renewables now generate more than 50% of our power.
• Curtailment fees paid to wind farms forced offline by grid bottlenecks—are soaring, adding billions to consumer and business costs.
• Grid delays stretch connection wait times into the 2030s and 2040s in some regions, deterring serious investment.

These barriers are not just academic. Already, we see UK-founded neoclouds like NScale choosing to deploy hundreds of megawatts of AI compute in Norway rather than at home. That is lost economic growth, lost jobs, and lost opportunity.

The solution requires urgent reform: data centres must be recognised as Energy Intensive Industries, renewable-heavy regions like Scotland should benefit from dedicated low-cost zones reducing curtailment fee’s, and government departments must work better, together to accelerate grid modernisation. Without this, Britain cannot compete for frontier AI and the AI Opportunities Action Plan will not work.

Copyright: Clarity Will Decide Investment Flows

The second fault line is copyright. Modern AI models are trained on vast datasets, much of which includes copyrighted text, images, and media. Jurisdictions that provide clarity and flexibility in their legal frameworks are attracting the lion’s share of AI investment.

• In the US, “fair use” provides a broadly permissive environment.
• In the EU, commercial training is allowed unless creators opt out.
• In the UK, commercial use of copyrighted material for training is largely prohibited.

This stricter stance may further comfort millionaires Elton John, Paul McCartney and parts of the creative sector, but it is already driving industrial AI developers to train their models abroad – which means infrastructure, the investment in that, the jobs that build it and the money that provides the economy all heads to the UK, Middle East and the Nordics. We’ve been told this may be resolved by 2029 – long after the global AI race has been run, and lost.

Frustratingly, the UK does not need to choose between supporting its creative industries and fostering AI innovation. With its unique strengths in both, it could set a global standard that protects rights while enabling progress. But that will only happen if government moves quickly to align copyright law with the EU’s opt-out approach and communicates a clear, investable framework to the AI ecosystem.

Infrastructure and Funding: Where Vision Must Meet Reality

The vision for unprecedented AI growth in the UK, and bustling, busy AI Growth Zones across the country will only materialise if there is demand. Investors need to know there will be long-term demand for 500MW-scale AIGZs. Demand for Culham, and potentially Teeside won’t just ‘occur’ unless it is under-written by the government. Commercial demand will only materialise if conditions – energy, copyright, regulation – are competitive. Without it, AIGZs risk becoming little more than branding exercises.

At the same time, we can’t promise to become the leading state for frontier AI if we don’t have the GPUs to do it. The government’s pledge to acquire 5,000 GPUs by 2030 is symbolic, but it pales against global benchmarks: Meta’s recent $10.5 billion investment secured around 50,000 Nvidia H100 GPUs in a single transaction.

Jensen put it bluntly when he visited London last time: “You can’t do machine learning, without a machine,” and unless Britain invests decisively in AI infrastructure, rhetoric will not translate into results.

It’s time we got real in the UK, stopped talking about possibilities, big ideas and lofty ambitions and got on with solving the real obstacles, the foundational building blocks that will genuinely make all of this happen. To that end, please download our AI Taker, not Maker whitepaper here: https://www.kaodata.com/taker-not-maker and if you’re equally frustrated, give me a shout and share your pain.