What does the UK’s £14bn AI ambition mean for innovation and start-ups?

“I’ve long advocated for the government to act as a customer for AI start-ups,” says Nigel Toon, founder and CEO of Graphcore. “If government bodies can become early adopters, it could be a game-changer for UK innovation.”

Toon was speaking about the government’s recent AI strategy announcement. Built on Entrepreneur First co-founder Matt Clifford’s AI Opportunities Action Plan, the strategy is about “laying the foundations to enable AI.” This means a focus on improving access to quality data through, among other things, a national data library, as well as developing “AI growth zones” and incentives for upskilling.

It is without doubt a bold idea that has broadly received warm praise. For Toon, already a veteran in what is still a nascent AI industry, the strategy is welcome news but of course, it is still early days.

“Governments have struggled to pull the data piece together in the past, and whether there can be some breakthroughs now will be quite interesting,” says Toon. “It’s good to see investment in compute infrastructure, but we need to ensure it benefits not just research clusters but also smaller enterprises.”

The proof of government policy is in the delivery and this government will be judged on how it manages the complexities of public and private interests. Regulation, investment, and actually improving AI talent pipelines will be challenging. But also how this idea of enabling researchers and enterprises of all sizes to benefit and not to just let this be the realm of the highest bidder will be key to driving growth.

“The challenge is proving that the government can be a reliable and engaged customer for start-ups and SMEs,” says Toon. “Smaller, more innovative companies might actually solve some problems more effectively than the traditional IT suppliers government has historically relied on.”

Toon mentions access to computing power, such as Isambard-AI, as also key. Based at the National Composites Centre near Bristol, Isambard-AI recently revealed its role in helping researchers develop new tests and treatments for a range of conditions including Alzheimer’s, emphysema and different types of cancers. 

As Professor Simon McIntosh-Smith, director of the Bristol Centre for Supercomputing and a leading expert in Higher Performance Computing (HPC), said recently, Isambard-AI is “already operational and with full production coming this year, it will play a central role in delivering a 20-fold increase in the UK’s AI capability within just a few years – faster and more cost-efficiently than commercial alternatives. This is a pivotal moment for AI in the UK.”

“The correlation between education and economic outcomes is becoming clearer. We need a highly educated workforce to think about AI application effectively. The government must ensure the right kind of talent flows through the system.”

Nigel Toon, Graphcore

For start-ups and researchers, access to cutting-edge infrastructure like Isambard-AI and Dawn in Cambridge could significantly level the playing field. While traditionally such resources have been the preserve of larger enterprises or academia, the AI Opportunities Action Plan opens the door for smaller players. “Access to advanced computing resources will not only accelerate innovation but also create opportunities for SMEs to compete on a global scale,” Toon notes.

This ties directly into the broader challenge of democratising AI, but the question remains whether the government can create pathways for start-ups to fully integrate into this ecosystem while maintaining an emphasis on innovation and inclusion.

“Translating this ambitious plan into reality will require careful planning and execution,” says Oliver King-Smith, CEO of Edinburgh-based SmartR AI. “Ensuring effective collaboration between government, industry, and academia will be critical.” King-Smith adds that while the plan mentions addressing security risks, “it’s important to ensure that ethical considerations are at the forefront of AI development and deployment. This includes issues like privacy, accountability, and the potential impact on employment.”

AI talent need apply here

While the UK faces stiff competition from other countries such as the US and China (not least since the launch of DeepSeek), any advantage will be delivered through innovation and the development of talent. But this is always easier said than done. UK education has long been a political football and suffered from chopping and changing policies. Our education system has to be backed, at all levels, to deliver on promises of increasing a skills-based, AI-savvy workforce. With this in mind, any increased investment in attracting talent, as well as upskilling opportunities, will be welcomed.

“The correlation between education and economic outcomes is becoming clearer,” says Toon. “We need a highly educated workforce to think about AI application effectively. The government must ensure the right kind of talent flows through the system.”

It’s a view supported by Dr Kjell Carlsson, head of AI strategy at Domino Data Lab. Carlsson is a former principal analyst at Forrester Research where he covered AI, ML, and data science. 

“The most valuable parts of the [government AI] plan focus on the most valuable resource in the AI era: human talent,” says Carlsson. “Attracting top global AI talent and making it easy for individuals to study, research, and launch AI ventures in the UK should be front and centre of the plan – not buried as a secondary priority.”

This was something that Matthew Forshaw, Senior Advisor for Skills at The Turing Institute highlighted recently. He cited two pieces of research, the first being that at least 80% of 2030’s predicted UK workforce are already in employment, and that a recent report by FutureDotNow and Lloyds Bank estimates that around 18% of the UK labour force (7.3 million people) lack the essential digital skills needed for the workplace.

Upskilling will clearly be crucial here, but that also means identifying the key skills that will be required to take full advantage of AI. It’s no good just saying we need to be more AI-literate. What does that really mean, in practical terms?

“We already have more AI engineers than Germany,” says King-Smith, “but at a per capita level, we want to become best in class and a new scholarship program is being developed (like Rhodes, Marshall, Fulbright) for domestic and international students to enhance the attractiveness of studying AI in the UK.”

National Centre for Universities and Business (NCUB) CEO Dr Joe Marshall supports this idea, saying that the UK requires “swift action” to develop the knowledge to continue to innovate AI, and the skills to harness these technologies.  

“Higher education is proven to be the most common pathway into AI careers and will likely remain so,” adds Marshall. “Universities are critically important to realise our future AI workforce.”

For Keri Gilder, CEO of Colt Technology Services, the government has “an unmissable opportunity” to address rather than reinforce gaps in areas such as skills and investment.

“Globally, research has proven that 71% of AI-skilled workers are men, compared to 29% [who are] women; and only one in five older workers have been offered AI skills training, as opposed to almost 50% of younger workers,” says Gilder. “The UK’s AI investment is predominantly clustered in London, and only 33% of AI funding is directed towards companies in growth and established stages.”

Gilder adds that these gaps will hinder the UK’s ability to innovate, expand and compete on a larger scale. But it also comes down to access. As Gilder suggests, this cannot be a postcode lottery. The AI Growth Zones are expected to reflect this, but while the first will be in Culham, Oxfordshire, home to the UK’s Atomic Energy Authority, we won’t know the other zones until the summer.

Marshall understandably suggests the zones “should complement robust regional clusters,” and “universities, as regional hubs of innovation and infrastructure, are essential to advancing AI infrastructure development.”

Ultimately what we want to see is action. As Julian David, CEO of techUK, says, “clearer timelines on AI infrastructure, streamlined visa pathways for talent, and strategic planning for semiconductor supply are essential to keep the UK competitive,” adding that “it is time to act, and at pace.”

Given the scope of the strategy and demands on resources, prioritising will be essential. Carlsson at Domino Data Lab suggests that to really get things moving, “resources earmarked for infrastructure would deliver greater value if redirected toward funding cutting-edge research, incubating AI start-ups, and crafting a regulatory framework that prioritises addressing real harms over hypothetical risks.”

Talent, regulation, computer power, regional access. If the strategy can lay the foundations, level the playing field for innovators and start-ups and deliver easy access to AI tools, it can transform and accelerate ideas. This in turn should attract investment. In short, this is a strategy to be welcomed – but the real proof will be in the delivery.