Two per cent. That’s how much the average UK business spends on AI as a proportion of revenue, according to a new study from PwC. The global leaders, companies in the top 20 per cent of AI-driven performance, spend five per cent. And they’re getting 15 per cent back. UK firms get 10.
You could look at those numbers and think the gap is manageable. A couple of percentage points here, a few there. But the study tells a more uncomfortable story when you dig into what’s actually happening with the money.
Only 27 per cent of UK businesses have redesigned their workflows to properly incorporate AI. The same proportion have updated their legacy IT systems to make it work. Which means roughly three quarters of British companies are doing what we’ve always done with new technology in this country: bolting it on top of what already exists and hoping for the best.
Anyone who’s worked in financial services for more than a few years will recognise this pattern instantly. We did the same thing with digital. The same thing with mobile. Adopt the shiny new capability, leave the underlying process untouched, then scratch our heads when the results disappoint.
The bolt-on approach has a predictable consequence, and PwC’s data captures it neatly. British businesses derived only 27 per cent of their revenue from products that didn’t exist three years ago. For global leaders, it was 43 per cent.
That’s where the real gap sits. Not in how much firms are spending, but in what they’re spending it on. Forty-eight per cent of UK businesses told PwC that efficiency and productivity were the main drivers of their AI experiments. Only 26 per cent cited revenue generation. In other words, most British firms are using AI to do the same things slightly cheaper. The companies pulling ahead globally are using it to build things that didn’t exist before.
Leigh Bates, PwC’s global risk AI leader, called the findings “a wake-up call.” He said businesses are being held back by legacy technology, underinvestment, and a tendency to think about AI as a cost-cutting tool rather than an engine of growth. The companies getting the biggest returns, he said, aren’t just doing more of the same. They’re reinventing their businesses.
Overall, the UK ranked 11th out of 19 countries surveyed, behind China (which topped the list), France, Germany, and Saudi Arabia. The US came in at 13th, which is an interesting detail. This isn’t simply a question of who has the deepest pockets. Something else is going on.
What’s going on, I think, is a mindset problem. And financial services is a useful lens through which to see it.
The mortgage industry runs on infrastructure that in many cases hasn’t been fundamentally rebuilt in over a decade. Origination platforms, CRM tools, core banking systems, all of it extended and patched but rarely rearchitected from the ground up. When firms in this sector adopt AI, they tend to do it at the edges. A chatbot here, some document automation there, maybe predictive analytics layered on top of existing data. Useful applications, every one of them. But they’re all cost-cutting plays. They’re all bolt-ons.
The firms getting outsized returns from AI globally are asking a different question. Not “how do we make our existing process cheaper?” but “what would we build if we were starting from scratch with this technology available?” That second question is harder. It means rethinking product design, customer journeys, operating models, sometimes the structure of the team itself. It’s uncomfortable work. But PwC’s data suggests, consistently, that it’s where the value actually lives.
There’s a timing dimension to this that matters. Bates made the point that the next 12 months will determine whether the UK becomes an AI leader or gets stuck in the middle of the pack. That sounds like consultancy rhetoric, but the logic underneath it is solid. The gap between leaders and followers in AI adoption isn’t closing. It’s widening. Companies that are investing now in genuine transformation, not just experimentation, are compounding their advantage quarter by quarter. The longer you wait to move from pilot to scale, the harder the gap becomes to close.
For mortgage firms, where regulatory complexity and legacy infrastructure already slow the pace of change, that window might be even tighter than PwC suggests. Consumer Duty has raised the bar on what firms need to demonstrate about how they serve customers. The Mortgage Charter has changed expectations around forbearance and communication. The regulatory direction of travel is towards more transparency, more personalisation, more responsiveness. AI can deliver all of those things, but only if it’s woven into the fabric of how a firm operates, not strapped to the outside.
The temptation is always to treat these studies as background noise. Another consultancy, another report, another set of statistics about how Britain needs to try harder. But the PwC numbers deserve more attention than that. The UK isn’t failing at AI. It’s underperforming. Spending enough to feel like it’s in the game, but not enough to actually change the score. Getting a return, but not the return that’s available.
The question Bates is really asking, underneath the diplomatic language, is whether British businesses have the appetite to do the hard, messy, expensive work of genuine transformation, or whether they’ll keep optimising what they’ve already got and hoping that’s enough.
For anyone running a mortgage business right now, that’s a question worth sitting with.
Source: The Times, 12 April 2026