Across government, a familiar pattern is emerging. Ministers announce new funding to address capability gaps – £750m for a national AI supercomputer, £27m for TechLocal skills delivery, £21m for hardware-level cyber defences, £16m for the Cyber Growth Action Plan.
Each announcement comes with ambitious targets:10 million workers trained in AI by 2030, critical infrastructure secured, digital transformation accelerated.
The instinct is understandable. Identify a problem, allocate resources, demonstrate action. But in complex systems, money solves only certain kinds of problems – and capability gaps in a post-industrial economy are rarely among them.
The industrial fallacy
Treasury thinking remains shaped by industrial-era assumptions. Add more looms, hire more weavers, produce more cloth. Within limits, inputs scale linearly to outputs. This logic works tolerably well for manufacturing physical goods with stable demand.
Modern government operates differently. The binding constraint is not capital but cognitive capacity – the skills, attention, and judgment of people who must design, implement, and operate complex systems. These do not scale linearly. You cannot double the number of qualified AI trainers by doubling the training budget. You cannot purchase ministerial attention or institutional focus.
The AI Skills Boost programme illustrates the gap between ambition and mechanism. Ten million workers trained by 2030 implies roughly 1.7 million people per year acquiring meaningful AI competence. Who will teach them? The UK already faces acute shortages of qualified AI and data science instructors. Funding more courses does not conjure the experts to deliver them. It creates competition for the same scarce talent pool, inflating costs without expanding real capacity.
In practice, this often means the same small group of subject-matter experts is asked to design curricula, train instructors, advise departments, and sit on assurance panels – all in parallel. The result is not faster delivery, but fragmented attention and slower progress across every programme drawing on that expertise.
When investment increases friction
Worse, large funding injections into constrained systems often generate their own overhead. Each new programme requires governance structures, reporting cycles, and accountability mechanisms. These consume the very leadership attention that is already the scarcest resource.
As I argued previously in these pages, initiative overload is not a resourcing problem but a cognitive one. Optimism bias launches programmes on best-case assumptions. Loss aversion prevents consolidation. Salience bias redirects leadership attention to whatever shouts loudest. New funding accelerates this cycle rather than breaking it.
The result is predictable. Organisations become busier but not more capable. Delivery teams juggle competing demands. The time between identifying a need and responding to it lengthens, not shortens – because every new initiative must navigate an increasingly crowded portfolio.
The constraint that cannot be bought out
Even when the government correctly identifies a bottleneck and directs investment toward it, institutional inertia limits the speed of response. Building new capacity takes time: recruitment, onboarding, and the months before new hires become fully effective. A department that secures funding today for AI security specialists may find that by the time those roles are filled and productive, the threat landscape has moved on – and the skills gap has simply shifted elsewhere.
This is not an argument against investment. It is an argument for recognising what investment can and cannot achieve. Capital can purchase equipment, infrastructure, and contracted services. It cannot purchase the years of experience that make an expert, the relationships that enable coordination, or the organisational focus that turns activity into outcomes.
Living within constraints
The alternative is not resignation but realism. Rather than assuming constraints can be funded away, effective organisations design around them.
A more pragmatic approach is to treat specialist expertise and leadership attention as throughput-limiting resources. Identify the genuine bottleneck. Subordinate everything else to maximising its throughput. Do not flood the system with work that will queue behind the constraint, creating noise without progress.
In government terms, this means treating leadership attention and specialist expertise as the finite resources they are. If a department has ten qualified people to deliver a capability, it should not commission fifty initiatives requiring that capability. Limiting work-in-progress is not a constraint on ambition but a precondition for delivery.
What this means for policy
None of this suggests that the investments announced in recent months are misguided. AI compute capacity matters. Cyber skills matter. Digital infrastructure matters. The question is whether funding announcements are accompanied by an honest assessment of the human and organisational capacity to absorb them.
This is not a uniquely British problem. Governments across developed economies face the same dynamic: ambitious funding commitments that outrun the capacity to deliver.
Ministers announcing new programmes might usefully ask: who specifically will deliver this, and what are they currently doing? If the answer involves the same overstretched teams and the same contested leadership attention, the funding may generate activity without outcomes – and taxpayers will rightly ask why.
The most important capability investment may not be the next programme but the discipline to stop, consolidate, and focus – so that the people and attention we have can operate at full effectiveness rather than perpetually context-switching between competing demands.
In a world that changes faster than large institutions can adapt, the question is not how much we can start – but whether we have preserved the capacity to respond when it genuinely counts.
Vsevolod Shabad researches and advises on cognitive bias, governance failures, and decision-making in complex organisations. His work draws on executive experience in critical infrastructure and technology across eight countries.