UK Research and Innovation (UKRI) is offering access to national-tier compute resources—specifically GPU hours—for AI-centred scientific research through the AIRR programme. Researchers can request 200,000–1,000,000 GPU hours to support large-scale projects in fields such as materials science, medical research, fusion, and engineering biology. This call provides compute only (no funding) and aims to accelerate AI-driven scientific discovery aligned with UK national missions.

Overview

The UK Research and Innovation (UKRI), under its AI for Science strategy and the AI Research Resource (AIRR) programme, provides advanced computational resources to support AI-centred scientific research. This call is the first mission-driven research opportunity aligned with the UK Compute Roadmap, enabling academia, industry, and independent researchers to access national GPU systems.

Why It Matters

This initiative strengthens UK leadership in AI-powered scientific innovation. It:
• Accelerates breakthroughs in high-priority scientific areas
• Supports the government’s national missions, including clean energy and health innovation
• Enables multidisciplinary, multi-sector research collaborations
• Provides access to large-scale compute that most institutions cannot supply internally
• Drives automated and autonomous scientific discovery through next-generation AI models

Key Focus Areas

Projects must align with AI for Science priorities, including:
• Materials science
• Nuclear fusion
• Medical research & health innovation
• Engineering biology
• Quantum technologies
• AI-driven scientific discovery & modelling
• Experimental, feasibility, and fundamental scientific research

Projects should contribute to at least one of the UK Government’s five missions:
• Economic growth
• A future-ready NHS
• Safer communities
• Expanding opportunity for all
• Clean energy and net-zero leadership

Available Compute Resources

• Total indicative pool: 8,000,000 GPU hours
• Each proposal may request: 200,000 to 1,000,000 GPU hours
• Systems available: Dawn and Isambard-AI
• Project duration: 6 months
• Note: This call provides compute access only—no financial funding is awarded

Who Is Eligible?

Eligibility is determined by UKRI’s standard research criteria.

Eligible Applicants

• Researchers across the UKRI remit (academia, industry, independent research)
• Applicants with a contract longer than the project duration
• For academic-led proposals: project leads must be at lecturer level or equivalent

Not Allowed

• Proposals seeking cash funding
• Projects unrelated to AI-centred scientific research
• Applications requesting compute beyond the allowed GPU-hour limits

How to Apply: Step-by-Step

1. Define the AI-centred research objective aligned with AI for Science priorities.

2. Choose the appropriate compute system (Dawn or Isambard-AI).

3. Prepare the research proposal, including scientific goals, expected impact, team composition, and compute justification.

4. Specify GPU-hour needs (within 200,000–1,000,000 hours).

5. Submit the application through the designated UKRI submission system.

6. Await review and allocation decisions.

Evaluation Criteria

Applications are assessed on:
• Scientific excellence and clarity of research aims
• Alignment with AI for Science strategy and UK national missions
• Potential for real-world impact and transformative outcomes
• Quality of team, partnerships, and multidisciplinary integration
• Technical feasibility and justification of GPU-hour usage
• Contribution to the development of new AI models or virtual scientific systems

Common Mistakes to Avoid

• Requesting compute hours without clear justification
• Proposing standard HPC computation instead of AI-centred research
• Weak alignment with priority areas or national missions
• Submitting a proposal with insufficient team expertise
• Failing to show how the project advances automated scientific discovery
• Assuming financial support is included (only compute is offered)

FAQ

1. Does this opportunity include funding?
No. It provides compute resources only.

2. Who can lead a project?
Applicants with contracts longer than the project period; academics must be lecturer level or equivalent.

3. What systems are available?
Dawn and Isambard-AI.

4. How long can the compute resources be used?
All GPU hours must be used within a six-month project window.

5. How many GPU hours can I request?
Between 200,000 and 1,000,000 hours per application.

6. Can industry researchers apply?
Yes. The opportunity is open across the UKRI remit, including industry and independent research.

7. What types of research are eligible?
AI-centred scientific research, including fundamental, experimental, feasibility, or industrial R&D.

Conclusion

The UKRI AI for Science and AIRR compute call offers a rare opportunity to access national-scale GPU resources essential for modern scientific breakthroughs. By supporting AI-driven discovery in priority domains and aligning with the UK’s strategic missions, the programme empowers researchers to push scientific boundaries and accelerate innovation across disciplines.