Britain's Digital Pothole: You Can't Build an AI Economy on Infrastructure That Can't Stream a Video Call
The UK ranks 57th globally for mobile network performance. That's not a tech problem. It's a strategy problem.
The UK ranks 57th globally for mobile network performance. That’s not a tech problem. It’s a strategy problem.
A Which? report published this weekend confirmed what anyone who has tried to take a video call on a British train already knows: the UK’s mobile internet coverage is worse than every EU country and every G7 peer. We rank 57th globally for network performance, 70th for download speeds, and 55th for the consistent quality needed for activities as basic as video calls, streaming, and gaming. More than a third of customers reported at least one problem with their mobile service in the past twelve months.[1]
Meanwhile, the Department for Science, Innovation and Technology (DSIT) talks about making Britain an AI superpower. UK Research and Innovation (UKRI) is committing billions to AI research and infrastructure. The Cabinet’s Digital and Technology Committee, chaired by the Prime Minister, is supposed to be coordinating this push.[2]
The contradiction is not subtle. And it is worth examining through a lens that most policy commentary ignores entirely: the psychology of why smart people keep making the same structural error.
The Optimism Bias Problem
Cognitive Bias Theory identifies a pattern that explains a great deal about infrastructure policy: optimism bias. We systematically overestimate the probability of positive outcomes and underestimate the likelihood of negative ones. It is why people start projects expecting them to finish on time and on budget, and why governments announce AI strategies without first checking whether the country can reliably load a webpage on a train between London and Manchester.
The planning fallacy is optimism bias’s operational cousin. It manifests as a preference for planning the exciting outcome (Britain as an AI leader) while underestimating the boring prerequisites (ubiquitous, reliable, fast connectivity). DSIT has published AI strategy after AI strategy, each one more ambitious than the last, and UKRI’s AI plan talks confidently about “world-class AI data and infrastructure”. None of them have begun with the sentence: “First, we need to fix the fact that our mobile networks are worse than Romania’s.”[3]
That is not a rhetorical flourish. According to Which?’s analysis of Opensignal data, the UK’s mobile infrastructure ranks behind not just the usual suspects but behind countries that most British policymakers would struggle to place on a map. Fifty-seventh in the world. For context, that puts us behind Bahrain, Kuwait, and North Macedonia. The country that helped invent the World Wide Web cannot reliably stream a video call on its own railway network.[4]
The Promotion Focus Trap
Regulatory Focus Theory distinguishes between two motivational orientations: promotion focus (concerned with gains, aspirations, and advancement) and prevention focus (concerned with security, reliability, and avoiding loss). The UK’s AI strategy is almost entirely promotion-focused. It speaks in the language of opportunity, leadership, and competitive advantage. It is exciting. It is forward-looking. It is, in the most precise psychological sense, aspirational.[5]
It is also built on sand.
Promotion focus is not inherently wrong. It drives innovation, attracts investment, and creates momentum. But when it operates without a corresponding prevention focus, it produces strategies that look impressive on paper and collapse on contact with reality. The reality is this: AI-mediated discovery, agentic systems, real-time machine learning, and the kind of AI infrastructure that actually transforms an economy all require one thing above everything else. They require connectivity. Not theoretical connectivity, not “coverage maps suggest” connectivity, not “5G is being rolled out” connectivity. Actual, reliable, fast connectivity that works when you need it, where you need it.
The UK does not have it.
Ofcom’s own Connected Nations reporting shows that while standalone 5G now reaches around 83 per cent of premises, and at least one operator claims 94 to 97 per cent outdoor 5G coverage, performance is uneven and reliability varies sharply by region. 5G coverage sits at roughly the mid-90s for outdoor availability from at least one mobile network operator, but that is not the same as consistent, high-quality service. 4G, the technology that was supposed to be mature by now, still does not reach everyone. These are not numbers that support an AI economy. They are numbers that support a country that has been talking about digital infrastructure for two decades and still cannot deliver it consistently across its own territory.[6]
The Infrastructure Gap Is the AI Gap
Here is the argument that most AI commentary misses: the real AI gap is not about models, talent, or research papers. The UK has all three in abundance. DSIT and UKRI are spending heavily. Private companies are building and deploying. The real AI gap is about whether the population can actually use the products.
AI is not a laboratory exercise. It is a utility. It is becoming the interface through which people search, decide, shop, learn, and work. If that interface requires reliable connectivity, and if a significant portion of the population cannot access reliable connectivity, then the AI strategy is not a technology strategy. It is a fantasy.
This is not a niche concern. More than a third of British mobile customers reported problems in the past year. That is not a statistic about inconvenience. It is a statistic about exclusion. Every dropped connection during a video GP appointment, every buffering delay during an online learning module, every failed transaction on a mobile payment system: these are the moments when the gap between the government’s AI narrative and the citizen’s lived reality becomes visible.[1]
The people most affected are not the ones reading AI strategy papers in Whitehall. They are the ones trying to use their phones in places where the signal does not reach. They are disproportionately in rural areas, in lower-income households, in the regions that the government’s “levelling up” rhetoric was supposed to prioritise. Ofcom’s maps and DSIT’s own evidence base make this clear. The digital pothole is not evenly distributed. It falls hardest on the people who can least afford to avoid it.[7]
The Institutions That Own This
If this is a structural failure, then specific institutions own pieces of it.
The Secretary of State for Science, Innovation and Technology, currently the Cabinet minister responsible for DSIT, owns the AI narrative and the compute roadmap. Ofcom, led by its Board and Chief Executive, owns the regulatory framework for mobile networks and the Connected Nations data that show just how uneven coverage and performance still are. The Digital and Technology Cabinet Committee, chaired by the Prime Minister and including the Chancellor and the Secretary of State for Science, Innovation and Technology, is supposed to align digital policy with economic strategy. The House of Commons Science, Innovation and Technology Committee and the Digital, Culture, Media and Sport Committee (and their successors in the new Parliament) scrutinise this space and have already warned, in earlier reports, that the UK would miss broadband and 5G targets.[8]
None of these bodies can plausibly say the problem is invisible. Ofcom’s own data show that standalone 5G coverage and real-world performance are diverging. DSIT’s calls for evidence on the mobile market explicitly acknowledge patchy coverage and uneven competition. Parliamentary committees have already written that the UK is likely to miss earlier gigabit broadband and 5G ambitions.[9] Yet the AI narrative continues almost unchanged.
Targets Without Alignment
The government’s flagship connectivity ambition is for all populated areas to have access to high-quality standalone 5G by 2030. Ofcom and DSIT are currently consulting on how to shape the mobile market to deliver that, and operators have publicly signed up to similar timelines.[10]
But 2030 is the wrong horizon.
DSIT’s AI Opportunities Action Plan and UKRI’s AI strategy are operating on a much shorter cycle. AI copilots, generative assistants, and real-time decision systems are being embedded into products and public services now. The NHS, local authorities, universities, and private firms are all running AI pilots that assume the underlying networks will hold.[11]
If the connectivity target sits five years behind the AI deployment curve, then the country is, by design, building advanced systems on top of a brittle base.
That is not strategy. It is wishful thinking.
What Needs To Happen, And Who Needs To Do It
This is not a call for another glossy AI prospectus. It is a call for specific people and committees to do their jobs differently.
The Secretary of State for Science, Innovation and Technology should treat connectivity as part of AI infrastructure, not as someone else’s problem. DSIT’s compute roadmap must be joined by a connectivity roadmap with the same level of detail, the same 2030 horizon, and the same Cabinet visibility. If the department is willing to fight for billions for supercomputers, it should be willing to fight for enforceable standards on the networks that citizens actually use.[12]
Ofcom should move from passive measurement to active enforcement. That means setting minimum performance standards — for example, for download speeds and latency on mainline rail routes — and making them licence conditions, as Germany has done. It means publishing more granular, route-level and region-level performance data, so that the Prime Minister, the Digital and Technology Committee, and Select Committees can no longer hide behind national averages.[13]
The Digital and Technology Cabinet Committee, chaired by the Prime Minister, should treat AI and connectivity as a single agenda. It already brings together the Chancellor, the Secretary of State for Science, Innovation and Technology, and other key ministers. It is the forum where someone can say, plainly: “We cannot claim AI leadership while ranking 57th globally for mobile performance.” That sentence needs to be said in that room.[14]
The Science, Innovation and Technology Committee and the Digital, Culture, Media and Sport Committee in the Commons should tie every AI evidence session back to the infrastructure that underpins it. When DSIT officials talk about AI Growth Zones and sovereign compute, the question should be simple: “Can a care worker in Wrexham or a student in Hull actually use these systems on their phone without the connection dropping?”[15]
If the answer is no, the hearing should not move on.
The Contradiction At The Heart Of UK AI Policy
The UK government wants to be an AI superpower. DSIT is building compute clusters. UKRI is funding advanced research. The Prime Minister chairs a Digital and Technology Committee that is meant to turn this into economic advantage.[16]
At the same time, Britain’s mobile infrastructure ranks 57th globally for performance and 55th for consistent quality. Coverage is patchy. Reliability is uneven. A significant share of citizens cannot depend on their connection for basic tasks, let alone continuous AI services.[1]
Those two realities cannot both be true in any meaningful sense.
Optimism bias explains how ministers and officials have allowed themselves to believe they can. The planning fallacy explains why they keep pushing the hard work of infrastructure into later phases. But cognitive bias is not a defence. It is a warning. If AI is going to be the new interface of the British economy, then it has to work wherever people actually are — on trains, in rural clinics, in small businesses, in homes without fibre — not just in ministerial speeches and demonstration labs.
Right now, it does not.
You cannot build an AI economy on infrastructure that cannot stream a video call.
David Chadderton is the creator of the STAR Framework and the author of three books: The STAR Framework: Rewriting the Rules of Consumer Engagement (NYC Big Book Award 2025), The STAR Operating System: Decode Mindset, Understand Motivation, Transform Human Behaviour, and Dear Algorithm, It’s Not Me, It’s You. He spent his twenties and thirties as a military aviator and instructor, studying how people make decisions when the stakes are highest. He now applies those principles as a Chief Marketing Officer, bringing behavioural science to performance marketing at scale. He writes about human behaviour, AI, and the psychology of decision-making on The Unoptimised Human.
Sources
[1] Signal Failure: The State of UK Mobile in 2026 — https://commsrisk.com/signal-failure-the-state-of-uk-mobile-in-2026/
[2] UKRI AI strategy makes bold choices where UK can lead the world — https://www.ukri.org/news/ukri-ai-strategy-makes-bold-choices-where-uk-can-lead-the-world/
[3] CP 1352 – UK Compute Roadmap — https://assets.publishing.service.gov.uk/media/6878b5dda52cca025ef5bd87/uk_compute_roadmap_web_accessible.pdf
[4] United Kingdom, January 2026, Mobile Network Experience Report — https://insights.opensignal.com/reports/2026/01/uk/mobile-network-experience
[5] AI Opportunities Action Plan Government Response — https://assets.publishing.service.gov.uk/media/678639913a9388161c5d2376/ai_opportunities_action_plan_government_repsonse.pdf
[6] Mobile Market Review: call for evidence — https://www.gov.uk/government/calls-for-evidence/mobile-market-review/mobile-market-review-call-for-evidence
[7] Best UK Mobile Network Coverage 2026 — https://blog.meaningfulplanet.co.uk/articles/uk-mobile-network-coverage-guide-signal-data-speeds
[8] List of Cabinet Committees and their membership — https://www.gov.uk/government/publications/list-of-cabinet-committees/list-of-cabinet-committees-and-their-membership
[9] UK government ‘likely to miss’ broadband and 5G targets — https://www.bbc.co.uk/news/technology-55403099
[10] The UK’s 5G rollout: a progress report — https://www.hamilton-barnes.com/resources/blog/the-uk-s-5g-rollout--a-progress-report/
[11] UK Government issues response to the AI Opportunities Action Plan — https://www.scl.org/uk-government-issues-response-to-the-ai-opportunities-action-plan/
[12] CP 1352 – UK Compute Roadmap — https://assets.publishing.service.gov.uk/media/6878b5dda52cca025ef5bd87/uk_compute_roadmap_web_accessible.pdf
[13] Mobile Market Review: call for evidence — https://www.gov.uk/government/calls-for-evidence/mobile-market-review/mobile-market-review-call-for-evidence
[14] Cabinet committees — https://www.instituteforgovernment.org.uk/explainer/cabinet-committees
[15] UK government ‘likely to miss’ broadband and 5G targets — https://www.bbc.co.uk/news/technology-55403099
[16] Ministers — https://www.gov.uk/government/ministers
The STAR Framework
If you enjoyed this essay, you'll find the full argument — and the framework behind it — in the book.