There are two ways to read a record funding number. One is that the market is broadening. The other is that a small number of very large deals have become big enough to make the aggregate look healthier than the underlying market.

The UK's first-half AI figures look much closer to the second explanation. Funding reached $9.6bn, or about £7.1bn, according to Tracxn data reported by UKTN. That was 360% higher than the same period a year earlier. Yet only 69 deals were tracked, down from 74 in H1 2025. The top five funded companies accounted for 84% of all capital.

London's 98% share is extraordinary even by British standards

The geographic concentration is sharper still. London companies accounted for 98% of the total, compared with 93% in the second half of 2025. Oxford was a distant second at $103m, followed by Cambridge at $18.3m, Edinburgh at $16.1m and Milton Keynes at $12m.

London has structural advantages that make a high share unsurprising. It combines international venture funds, large corporate customers, universities within commuting distance and one of Europe's deepest pools of software and financial talent. British Business Review's guide to London's biggest technology employers shows how unusually dense that labour market has become. The question is whether 98% represents a productive cluster effect or a capital-allocation imbalance that leaves viable companies elsewhere struggling for attention.

Big rounds are doing more work than deal count

The funding surge was driven by companies including Isomorphic Labs, Nscale and Wayve. Together, those three accounted for $5.3bn. That is a sign of confidence in a small group of ambitious British AI businesses, but it is not the same thing as a fourfold improvement in conditions for the median seed-stage founder.

For investors, concentration can be rational. Frontier-model research, autonomous driving and AI infrastructure are capital-intensive. A company that needs expensive compute, specialised hardware or large research teams will absorb more money than a conventional software startup. But the gap between funding growth and deal-count growth should stop policymakers from using the headline total as evidence that every part of the ecosystem is equally well financed.

The regional policy challenge is customer access, not just local funds

The usual answer to regional concentration is to create more local investment vehicles. Capital matters, but AI startups also need early customers, experienced technical hires and repeat founders. London has all three in unusually high density.

A more durable regional strategy would connect companies in Cambridge, Oxford, Bristol, Manchester, Edinburgh and other clusters to national procurement, corporate buyers and specialist compute rather than trying to reproduce London's venture market city by city. The government's sovereign-compute programme can help if access is genuinely national. Public procurement can help if smaller firms outside London can compete for it without moving their commercial operation south.

What the second half needs to prove

The most encouraging H2 outcome would not necessarily be another record headline. It would be a broader distribution of rounds by stage and geography, with more companies raising meaningful Series A and Series B capital rather than a handful of mega-rounds lifting the aggregate.

London's dominance is a competitive asset for Britain. It becomes a vulnerability only if the rest of the country's AI economy cannot plug into it. The first-half numbers say the capital is here. They do not yet show that it is reaching the whole market.