The Government’s 2025 AI Opportunities Action Plan is an exciting vision. It is exactly what the country needs – riding a strong technology wave to higher economic growth and better vital public services, such as the NHS. It has changed the game for mobile infrastructure strategy: lifting the bar on mobile data speeds is no longer strategically important — what matters now is lifting the bar on network reliability. On this measure, the UK’s mobile infrastructure is in a poor state. Ofcom must pivot from managing ever-rising video data to safeguarding the value of lower-volume but critical AI traffic. The AI Action Plan itself identifies the value as — driving economic growth, improving healthcare, and transforming how citizens interact with government. In an AI-driven society, the reliability of these data flows will be as vital as the services themselves.

The reliability problem

The success of the AI Opportunities Action Plan depends on a mobile infrastructure delivering time-critical, high-value data flows to and from our smartphones — reliably, instantly, and wherever we are. The reality is starkly different: UK 5G networks are over five times thinner than the world’s best-in-class 5G network (in China, a leading strategic competitor); 70% of the country has wafer-thin mobile data capacity; transport corridors remain full of coverage gaps; minor roads lack coverage, and up to a million buildings block indoor mobile signals.

Compounding the problem, our networks were never designed to handle time-sensitive AI traffic — where late or inconsistent data can erase economic value or compromise safety.

Why this matters now

We are at the hardest point in 20 years to upgrade the reliability of our mobile infrastructure:

• Ofcom’s auction pipeline of usable mobile spectrum is now exhausted — closing the low-cost route to increasing network data capacity.
• Public policy requirements for security, resilience, and net-zero are increasing demands on a finite investment pool.
• There is no realistic prospect of taxpayer subsidy to fill market failure gaps.
• The global vendor-funded R&D model, which has powered four generations of mobile advances, is collapsing — this could particularly expose the UK.

With a strategic intent that only endorses the regulatory status quo, the UK will enter the AI era with a network unfit for purpose and a barrier to time-critical AI innovations.

The cost of finding this out the hard way will be far greater than acting now.