Research infrastructure: why confidence, not technology is the real challenge
UK universities face growing uncertainty in research infrastructure planning, where confidence in decision-making matters as much as the technology choices themselves.
Ask a university leader about their biggest research infrastructure challenge, and the answer is rarely which GPU cluster to buy or whether to move to the cloud. More often, it is whether they can justify major investment decisions when future demand, funding priorities, technology requirements and operating costs remain difficult to predict with confidence.
That was the clearest message from an event hosted by our research and innovation strategy forum, a group of pro-vice chancellors for research and innovation from universities across the UK. The event brought together senior research leaders from 21 higher education institutions, and included presentations from University College London, Queen Mary University of London, the University of Warwick, Imperial College London and the University of Oxford.
The discussion confirmed something that does not always come through in policy discussions about compute and AI infrastructure: the sector’s primary challenge is not technology selection. It is decision-making under uncertainty, a theme we are exploring further through our research infrastructure survey.
Decision-making under uncertainty
What struck me most was how many institutions were already well advanced in their decision-making. Even among those with mature digital research infrastructure strategies, there was a clear recognition that provision must be continually reassessed as research requirements, sustainability objectives, technology innovation and financial constraints evolve.
Research infrastructure planning has become structurally more complex. Institutions are no longer simply deciding which hardware to procure. They are balancing power, estates, sustainability, funding and workforce constraints simultaneously.
The constraints have changed
AI workloads are accelerating this shift, pushing infrastructure demands beyond the assumptions on which many university facilities were designed. In some cases, the limiting factor is no longer access to compute but whether campus infrastructure can provide the power, cooling and physical space and cost stability needed to support it.
Research infrastructure planning has become structurally more complex.
As a result, infrastructure decisions are now as much about estates, sustainability, finance and governance as they are about technology. They require a broader range of stakeholders and a different set of conversations than they did even five years ago.
No single blueprint
One of the most valuable aspects of the discussion was hearing how differently institutions are approaching the same underlying problem.
Some are investing in local facilities to maintain control and meet specialist research requirements. Others are pursuing colocation, commercial partnerships or cloud services. Many are combining approaches.
Hybrid provision emerged as the dominant direction of travel, reflecting institutions' efforts to balance flexibility, cost, sovereignty and sustainability across different workload types. Presentations from Professor James Hetherington (UCL), Professor Jonathan Hays (Queen Mary University of London), Dr Matt Ismail (University of Warwick), Dr Andrew Richards (Imperial College London) and Professor Brian Marsden (University of Oxford) demonstrated how institutions are approaching similar challenges in different ways, encompassing workforce development, governance, federation, sustainability and future infrastructure provision.
Hybrid provision emerged as the dominant direction of travel.
Despite these different approaches, the challenges discussed were remarkably consistent: forecasting future demand, comparing costs across provision models, attracting and retaining specialist skills, and making the case for long-term investment in an environment shaped by short-term financial pressures.
Turning infrastructure into research capability
Institutions also need onboarding support, researcher capability, governance frameworks and operating models to use those resources effectively. The gap between infrastructure availability and infrastructure adoption is real, and it is not primarily a technical problem. Closing it requires sustained investment in people, practice and shared understanding across the sector.
These conversations should not be happening in isolation
What the event demonstrated more than anything else was how much institutions have to learn from one another, and how rarely they get the opportunity to do so.
Institutions across the UK are grappling with similar questions around demand forecasting, investment, skills and sustainability, yet many are working through them independently. The result is a collective cost: slower learning, weaker benchmarking and missed opportunities to build the evidence base needed by institutions, funders and policymakers.
Professor Nick Brook, pro vice-chancellor for research and innovation at Manchester Metropolitan University and Chair of our research and innovation strategy forum, reflected:
“What came through strongly at the forum is that while our institutional starting points differ, the challenges we face are remarkably shared — and so too is the opportunity to think more collectively about how we provide, share and sustain research computing and data infrastructure across the UK. Perhaps most strikingly, this is not simply a question of facilities and investment. It is equally about people: the skills, capabilities and capacity we need across the sector to design, operate and fully realise the value of that infrastructure."
Building that evidence base is one reason we launched our survey, research compute and data centres: strategic choices for UK universities.
By gathering insight from a wider range of institutions, we aim to develop a clearer picture of research infrastructure planning, investment priorities and future demand across the sector.
The research and innovation strategy forum plays an important role in this work. By creating space for open, cross-sector discussion, it enables institutions to share experience, challenge assumptions and build a more informed understanding of the choices ahead.
Universities are making decisions now that will define their research infrastructure for a generation. The challenge is not a lack of technology options. It is how to make high-consequence decisions with sufficient confidence in future demand, costs, skills and funding conditions.
The challenge is not a lack of technology options.
Given the complexity and long-term implications of these decisions, many institutions are seeking opportunities to compare approaches, share experiences and learn from emerging practice across the sector. In support of this learning, we are extending the survey until September 2026 and inviting more pro vice-chancellors for research and innovation to join the research and innovation strategy forum.
This builds on a growing body of sector work, including our published mapping federation journeys and ongoing discussions about shared and federated approaches to research compute and data infrastructure. Together, these activities reflect a broader recognition that many institutions are grappling with similar challenges and that greater coordination may help accelerate learning, reduce duplication and improve the effectiveness of future investments.
While several common themes emerged from the forum discussion, we should be cautious about assuming they represent the views or priorities of the entire sector. Extending the survey will help test these findings with a broader group of institutions and identify where perspectives diverge as well as where they align.
If institutions are making some of the most consequential infrastructure decisions in a generation, research leaders need stronger evidence, broader perspectives and more opportunities to learn from shared experience.
Find out more
- Contact your relationship manager to join our research and innovation strategy forum
- Take part in the research infrastructure survey
- Explore our research communities
- Why trusted research environments should be at the centre of digital strategies to fuel innovation, sustainability and growth
- Improving the reliability of research management data across the sector
About the author
Sara is head of higher education research at Jisc. She works with universities, funders and policymakers to explore the future of digital research infrastructure, helping the sector navigate emerging challenges and opportunities across research computing, data, AI and innovation. Her work focuses on connecting institutional priorities with national research and technology developments to support a more sustainable, collaborative and resilient research ecosystem.