Event Roundup: Unlocking UK leadership in next-generation compute hardware
The UK government has committed up to £2 billion by 2030 to build a world-class compute ecosystem. This closed-door roundtable convened innovators, policymakers, and industry leaders to explore how the UK can capitalise on its strengths in advanced Semiconductor Design, Quantum, Photonics, and Neuromorphic technologies to secure long-term leadership in Frontier Compute.
The session addressed four core questions:
-
Making the strategic case: positioning Frontier Compute as a driver of economic growth, innovation, and national resilience.
-
Identifying the gaps: investment, manufacturing capability, supply chains, and workforce barriers.
-
The convergence opportunity: how the intersection of emerging hardware technologies represents a step-change in capability.
-
The approach: the respective roles of government and industry in enabling development and deployment.
Making the Case
Participants were broadly aligned that the UK cannot and should not attempt to lead in every aspect of hardware underpinning frontier compute technologies. The key question is where to focus. A counterpoint was raised that the UK should not adopt a defeatist attitude, and the starting position should be that the UK can win in any market it chooses. The real constraint is cultural, not technical.
The broad consensus agreed that it is too early to 'pick winners' and rule out any space. As sectors emerge and mature, the UK can be more selective, but spreading resources too thinly too early is the greater risk.
Sovereignty was identified as a primary driver; the UK must control some of its own hardware infrastructure. This is a matter of national resilience. Cyber resilience was cited as a compelling strategic case: digital security by design would address 70% of cyber resilience hits through new chip architectures, which could address the UK's cyber vulnerability exposure costing an estimated £9.2 trillion annually, as illustrated by recent attacks on JLR and M&S.
Participants also highlighted the importance of being clear about what makes these technologies different, as this supports making the case. For example, photonic chips can solve the energy cost challenge for AI infrastructure given their improved energy efficiency, as a result of using light rather than electrons to process data.
Why Convergence Matters Now
Several participants argued that the convergence of photonics, quantum, neuromorphic, and semiconductor technologies represents a fundamental shift, not just an incremental improvement. The UK has a potential window of 3–5 years before dominant players establish lock-in and set standards. Once a dominant player establishes standards and builds the technology stack, catching up becomes structurally very difficult.
Early leadership in a niche enables standard-setting, ecosystem-building, and the creation of sustainable competitive advantage. Traditional machine learning is likely past its peak (arguably since 2023). The current frontier is 3D chips, multi-chip systems, and optical computing. The UK needs to orient investment toward these emerging paradigms, not defend existing ones.
Gaps and Structural Barriers
Investment
The UK is investing less than one-eighth of what the US is committing to this space. £2 billion by 2030 is a welcome signal but described by several participants as insufficient, especially given the US, China, and EU are committing orders of magnitude more. For example, the UK was once a genuine global leader in supercomputers, but Germany is now taking over due to increased funding. The ambition should be to use the £2 billion to unlock £20 billion or £200 billion in private investment, leveraging public money as a catalyst, not a substitute.
The majority of leading US frontier compute investment is private-sector-led. The UK needs to find mechanisms that crowd in private capital rather than substituting for it.
Manufacturing and Industrial Base
Manufacturing was described in the session as anchoring, meaning that once capabilities are established, they tend to stay. Southampton was cited as an example: 18 optical fibre cable manufacturers clustered around the university, sustained by shared infrastructure and a pilot line that companies return to repeatedly. This model should be replicated.
Seagate's Northern Ireland site was raised as an example of where manufacturing capability could be lost, with this being the company's most expensive global location, making further investment difficult to justify.
Workforce and Skills
A long-term skills pipeline is essential and cannot be built quickly. There is a global shortage of relevant technical expertise. CDTs (Centres for Doctoral Training) in neuromorphic computing and related fields need to be identified and funded now. Participants expressed the need to look ahead and build the pipeline now that to meet demand in years to come.
Commercialisation
Infrastructure for prototyping and academic-industry collaboration is underdeveloped. The 'missing middle' between research and commercial deployment needs targeted support. Several participants noted that the UK has world-class academics and strong design capabilities, but consistently fails to scale and commercialise.
What Government Should Do
-
Intervene to correct market failure and ensure additionality -public investment should only go where it unlocks things that would not otherwise happen, and where it crowds in private capital.
-
Avoid prescribing technology choices, the role of government is to define the problem to be solved (e.g. speed of compute, energy efficiency, cyber resilience) and fund solutions, not pick winners from the top.
-
Shorten planning horizons in periods of high uncertainty: the ARIA model (here is the problem; now develop the solution) is the right framework for this moment.
-
Make a long-term commitment: the workforce and infrastructure needed for frontier compute cannot be assembled quickly. Government needs to signal multi-year commitment to attract the investment and talent required.
-
Use the AI Hardware Plan and the Compute Roadmap as vehicles to make the case across government, these have attention from the Chancellor and Secretary of State and should be used to drive coordinated action.
-
Government procurement should be leveraged as a demand signal particularly in defence, health, and financial services to pull through frontier compute capability.
-
Reduce energy costs to make the UK a competitive location for hardware manufacturing and data centre investment.
What Industry Should Do
-
Move beyond pilots: Industry needs to build the case for scaling, not just demonstrating.
-
Engage vertically: have specific, detailed conversations with defence, financial services, and health about application needs and build the hardware stack, not just focusing on the tech, but also the end use applications.
-
Build coalitions: every hardware company needs software partners. These partnerships need to be more actively brokered.
-
Avoid the language of 'niches' in public discourse, 'high-potential' or 'high-growth' technologies is more effective for building political and investor support.
Roles and Governance
-
The UK Semiconductor Centre will create strategic direction to feed into what industry needs, this was cited as a positive structural development. Innovate UK is shifting its focus toward growth-oriented, high-potential companies. This would represents a potential step change in support for scale-up. Government does not have the technical depth to make hardware architecture choices - industry needs to provide that intelligence clearly and accessibly, raising questions around the dialogue between government and industry.
This event is part of techUK's Frontier Compute focus. Find out more here.
Events
Insights
Authors
Elis Thomas
Elis joined techUK in December 2023 as a Programme Manager for Tech and Innovation, focusing on AI, Semiconductors and Digital ID.
Ella Shuter
Ella joined techUK in July 2025 as Junior Programme Manager for Emerging Technologies.
