The UK may have missed the AI frontier. Can it hold its advantage in quantum?

A session of Parliament’s Joint Committee on the National Security Strategy on 7 September offered an unusually useful comparison between two technologies that are often bundled together under the heading of “frontier”.

They are, of course, at very different stages of development. But that was partly what made the discussion interesting. The evidence on artificial intelligence was notably downbeat about the UK’s ability to compete directly at the technological frontier. The evidence on quantum was rather different.

Professor Kenneth Payne, professor of strategy at King’s College London, described an AI frontier that is “accelerating away”, driven by levels of investment, compute and infrastructure that are increasingly difficult for countries such as the UK to match.

“If you want to be at the frontier, the frontier is accelerating away from you,” he told the committee, adding that his “slightly pessimistic take” was that Britain had little prospect of catching it.

That does not mean the UK cannot extract considerable value from AI. Payne was clear that adoption matters, as does the ingenuity with which existing models are put to work. But the question is one of dependence. If access to the most capable models sits elsewhere, what does sovereignty mean in practice?

George Balston, strategy adviser at the Alan Turing Institute and co-founder of frontier AI safety organisation AVERI, drew attention to the UK’s relatively limited inference capacity, high energy costs and dependence on overseas infrastructure.

“We come up short across the stack,” he said.

A new hope

Quantum, by contrast, presents a more complicated and, for the UK, more encouraging picture.

Dr Michael Cuthbert, director of the National Quantum Computing Centre, argued that while quantum computing remains at a much earlier stage of development than AI, Britain has a stronger position across more of the underlying technology.

Global investment is dominated by the US and China, but Cuthbert described the UK as a strong third player, supported by more than a decade of sustained public investment and a research base that remains internationally competitive.

There are obvious gaps. Device fabrication is one. Much of the UK’s clean-room capability is still geared towards academic research rather than industrial-scale manufacture. Access to critical materials may become another. But the areas of strength are significant. This includes photonics, specialist manufacturing, quantum software, error correction, algorithms, and a substantial academic base.

“We have a much stronger grip on the quantum stack,” Cuthbert told the committee – but added that this position is fragile and needs to be maintained.

The question is whether the UK’s current scientific and technical strengths can be converted into something more durable before other countries pull further ahead. We may be good at quantum now, but that’s certainly no guarantee for future viability.

Cuthbert described a shift from government acting primarily as a funder of quantum technology to becoming a user of it. Early work is already taking place across defence and national security, including navigation in GPS-denied environments, work with GCHQ, and proof-of-concept projects involving defence primes.

That may prove more important than it first appears. Deep technologies rarely become economically significant simply because the research is excellent. They need customers, applications, procurement routes, engineers, supply chains, and people inside large organisations who understand what the technology can and cannot do.

Cuthbert suggested that some of these early projects are as much about learning as validation. They familiarise organisations with quantum technologies, build internal capability, and connect commercial users with academic expertise. In other words, the challenge is increasingly about working out where quantum is useful.

Another theme running through the session was that quantum is unlikely to develop in isolation.

Asked about future cybersecurity risks, Cuthbert suggested that the more plausible near-term concern is not necessarily a criminal organisation somehow acquiring its own cryptographically relevant quantum computer. Quantum capability might instead be used to enhance other forms of computation, including machine learning and high-performance computing.

The UK’s strength in algorithms becomes particularly interesting in that context.

“It will not be one or the other,” he said. “It will be in tandem.”

The future of quantum computing may be less about standalone machines replacing classical computing, and more about quantum processors sitting alongside high-performance computing and AI, each used for the problems to which they are best suited.

It also raises a question about where value sits. Building the largest quantum computer is one form of advantage, but it is not the only one. Algorithms, software, error correction, systems integration, and expertise within specific industries may prove just as important.

But what happens if quantum computers do not scale over the next decade exactly as current industry roadmaps assume?

Europe already has considerable strength in algorithm development. Rather than relying entirely on larger and larger machines, Cuthbert argued, there may be more value in finding better ways to use more modest systems.

People power

It’s about being honest and building real strength where the UK knows that it can sustain interest. The same could be true of skills. Cuthbert estimated that around 2,500 people currently work across quantum technologies in the UK, but believes the workforce may need to grow by at least a factor of ten over the coming decade. The conventional response might be to call for many more quantum physics PhDs. He argued that this would be far too narrow.

“Industry will scale on the back of engineering,” he said.

That means drawing people in from computing, systems engineering, electronics, and other adjacent disciplines, including experienced engineers who may already have spent 10 or 20 years elsewhere in industry.

It’s as much about image as substance. Quantum still carries a degree of mythology around it, which can make the field appear more closed than it really is. If the sector is to grow at the speed envisaged, some of those boundaries will need to be broken down.

And one sector which could provide some urgency to this is cybersecurity.

The National Cyber Security Centre has set out a timetable for migration towards post-quantum cryptography, with milestones running to 2035. Cuthbert was clear that organisations should not read that date as a convenient point at which to start thinking about the problem.

“In my view, it is much too late to wait until 2035,” he said.

Changing cryptographic infrastructure across government and industry is itself a long and complicated process. At the same time, quantum companies need to think about conventional cybersecurity throughout their own technology stacks well before cryptographically relevant machines emerge.

What happens next will depend on whether the UK can turn research into adoption, connect quantum effectively with AI and high-performance computing, bring a much wider engineering workforce into the sector, strengthen weak points in its supply chain, and create customers willing to experiment with the technology before every use case has been neatly resolved.

The committee was examining AI and quantum through the lens of national security. The UK is not alone in this, but it is better placed than most to forge ahead. Holding that position will however take sustained investment and faster decision making, neither of which were forthcoming in the early days of AI. Have lessons been learned?