The UK government just published its one-year scorecard on the AI Opportunities Action Plan. Thirty-eight out of fifty commitments met, billions in private investment, a new supercomputer running in Bristol. By any policy measure, it is working. So why are 97% of UK businesses still reporting an AI skills gap? That disconnect - between what the government has built and what most organisations are actually doing with it - is the most important AI story in Britain right now.
What the AI Action Plan has actually delivered
When the government published its AI Opportunities Action Plan: One Year On in January 2026, the numbers were genuinely impressive. This is not spin – it is a legitimate policy delivery story.
Thirty-eight of fifty commitments met in the first twelve months. Here is what that looks like in practice:
The Isambard-AI supercomputer went live at Bristol University in July 2025 – one of the most powerful AI systems in the world
£28.2 billion in private investment flowed through five AI Growth Zones, creating over 15,000 jobs
A Sovereign AI Unit was established with up to £500 million behind it
More than one million free AI courses were delivered – ahead of the government’s own schedule
The NHS is now AI-assisted on 2.4 million scans, and stroke treatment time has halved in participating hospitals
You can track the live progress at the official AI Action Plan delivery dashboard. The supply side – compute, investment, infrastructure, regulation – is largely sorted. Britain has built the environment in which AI can thrive.
And yet.
The £140 billion problem hiding in the scorecard
The entire point of the AI Action Plan is a single number: the OECD estimates that AI adoption could lift UK productivity by 0.4 to 1.3 percentage points – worth between £55 billion and £140 billion of gross value added by 2030.
That is the prize. And here is the critical thing about it: not one of the 38 completed commitments delivers it directly.
Supercomputers, Growth Zones, a Sovereign AI Unit – these are the supply side. They create the conditions for AI to be used. The productivity dividend only materialises when firms across the economy actually change how they work. That is a demand-side problem, and as Resultsense noted in their analysis published yesterday, it is exactly where the opportunity is currently leaking away.
The numbers that matter most
Only about 25% of UK firms are using AI at all – whole economy average, ONS/DSIT data, December 2025
Even in the most AI-advanced sector (professional and business services), adoption sits at 43.4%
Around 75% of firms are not ready on the core enablers – clean data, process integration, monitoring – needed to run AI properly
Just 16% of UK firms have made what counts as a strategic AI deployment
These are not numbers from a country about to collect a £140 billion productivity dividend. They are numbers from a country that has built an excellent runway and is still debating whether to take off.
Why 97% of UK businesses still have an AI skills gap
Here is a stat worth sitting with: 97% of UK businesses report having an AI skills gap. Over 60% cite insufficient internal skills as their primary barrier to AI adoption. These figures exist alongside all the government’s impressive supply-side wins – and they are not contradictory. They are the same story told from the other end.
The barrier is not access to AI tools. ChatGPT, Copilot, Gemini, Claude – all of them are available to any employee today, most for free. The barrier is that people do not know how to use them in a way that actually changes their work. They have the tools but not the skills. The organisation has the capability available to it but not embedded in it.
Three reasons the skills gap persists
Training has been patchy. The government has delivered over a million free AI courses, but most are introductory. Moving from awareness to practical, daily use requires structured, role-relevant training – which most employers have not provided.
Organisations have not redesigned work. The productivity dividend comes from changing a process, not just using a tool. Most businesses have given employees access to AI without changing anything about how work is structured around it.
The SME tail is barely started. Large firms are experimenting; smaller businesses have mostly not begun. The productivity gap between AI-fluent large firms and AI-absent SMEs is widening every month that passes without intervention.
This is what the Resultsense analysis describes as Britain’s recurring pattern: strong on invention and infrastructure, weak on diffusion. We build the breakthrough, then struggle to roll it out.
The supply side is done. The demand side is on you.
There is a useful distinction between two types of AI problem. The first is an access problem – not enough compute, too much regulation, not enough investment. Britain has largely solved that. The second is an adoption problem – organisations that have access but are not using it to change anything. That is where Britain is stuck.
The government cannot solve the adoption problem for you. It can fund training programmes, it can designate Growth Zones, it can run AI pilots in the NHS. But it cannot sit inside your organisation and redesign your workflows. It cannot upskill your team. It cannot make the decision to move from an AI pilot that is interesting to an AI deployment that saves you money.
What separates businesses that are winning with AI from those that are not
They have trained their people – not just once, not just theoretically, but in the specific tools they use for the specific tasks they do every day
They have redesigned at least one workflow end-to-end around AI, rather than just adding AI on top of existing processes
They measure AI by outcomes – time saved, cost reduced, quality improved – not by how many tools they have access to
They treat AI as an operations question, not an IT question
The gap between a business doing all of this and one that has given people access to Copilot and hoped for the best is enormous. And it is widening. The £78 billion of unrealised AI value estimated to sit in the UK SME sector alone is not a fixed number – it is leaking to the businesses that are moving faster.
What your business should do right now
This is not complicated, but it does require deliberate action. The businesses that will collect the productivity dividend are the ones that treat AI as an operations priority right now, not a technology experiment for later.
Here are three things to do this week:
Audit what your team is actually doing with AI. Not what tools they have access to – what they are using, how often, for what tasks. Most organisations have no idea. Find out, and you immediately know where the skills gap lives.
Pick one workflow and take it all the way. Not a pilot, not an experiment – one real process, properly redesigned around AI, measured, and running in production. That single workflow will teach you more than ten pilots.
Train people properly. Not a one-hour webinar about what AI is. Role-relevant, practical training that covers the specific tools they will use on the specific tasks they do. The ivee AI Masterclass series is a practical place to start – structured learning across the major AI tools, designed for people who need to use AI at work, not people who want to understand AI in theory.
The government has done the hard work of building the environment. The question now is whether your organisation picks up the opportunity it has been handed.
FAQs: UK AI Action Plan and workforce AI
What is the UK AI Opportunities Action Plan?
The UK AI Opportunities Action Plan is a government strategy launched in January 2025, setting out 50 specific commitments on compute, investment, regulation and skills. As of January 2026, 38 of those 50 had been met – including a twentyfold supercomputing expansion and £28.2 billion in private investment via AI Growth Zones.
What is the AI productivity dividend for the UK?
The OECD estimates AI adoption could lift UK productivity by 0.4 to 1.3 percentage points, worth between £55 billion and £140 billion in gross value added by 2030. But this only happens if businesses actually change how they work – building data centres and passing regulations does not book the number.
Why do so many UK businesses still have an AI skills gap?
97% of UK businesses report an AI skills gap, with over 60% citing insufficient internal skills as their main barrier. The problem is not access – anyone can use ChatGPT today. It is that people have not been trained to use AI effectively in real work contexts, and most organisations have not redesigned any processes around it.
What is the difference between AI adoption and AI in production?
Adoption means a business has tried a tool or run a pilot. Production means a real workflow has been rebuilt around AI and it is saving time or money. Most UK businesses are at the trial stage. The productivity dividend only comes from production – and around three in four firms are not ready on the data, process and monitoring needed to get there.
How can my team start using AI properly?
The most practical starting point is structured, role-relevant training – not a one-hour introduction to what AI is. ivee’s Core AI Programme covers ChatGPT, Claude, Copilot and more in the context of real workplace tasks. No coding required. The goal is to move from curiosity to daily, practical use as fast as possible.
The government has done its part. Now it is yours.
Britain has spent a year proving it can build the supply side of AI. The supercomputers are running, the investment is flowing, and 38 of 50 policy commitments are ticked off. What the scorecard cannot tick off is what happens inside your organisation.
ivee works with organisations who are serious at building AI capability throughout the organisation. Get in touch to discuss what a custom AI training programme might look like for your team.
