UK Innovators Partner To Explore AI In Scientific Research
Locai Labs Limited and First Light Fusion have signed a Memorandum of Understanding (MoU) to test how UK-built foundational AI can support high energy density physics and inertial fusion energy research. The work centers on improving technical workflows and assessing where agentic AI can boost machine learning-driven simulation.
Partnership scope
Under this non-binding agreement, Locai Labs' models will be evaluated against First Light Fusion's R&D needs, with a clear target: faster iteration from idea to insight to implementation. Initial areas include code development support and agentic systems that coordinate complex simulation tasks.
All AI processing will run on First Light Fusion's secure, air-gapped high-performance computing infrastructure in Oxford. This setup keeps sensitive data on-prem and under strict control, a key requirement for many labs working with proprietary physics and IP. For context on air-gapping practices, see the UK National Cyber Security Centre's guidance here.
Why this matters for scientists
Running a UK sovereign AI model inside an air-gapped HPC environment tackles data residency, confidentiality, and reproducibility head-on. If successful, the approach could reduce cycle times in high energy density physics and inertial fusion-where simulation throughput, code reliability, and experiment tracking often bottleneck progress.
Leadership from both companies emphasized practical outcomes: accelerating UK science and industrialisation, strengthening fusion research, and improving day-to-day efficiency across technical and operational workflows.
What the trial will evaluate
- Code development support for simulation teams, with guardrails suited to on-prem HPC workflows.
- Agentic AI to coordinate ML-assisted simulation tasks and increase throughput.
- Productivity gains in document management, internal communications, and administrative tasks.
- Security, auditability, and operational fit for air-gapped, regulated environments.
Statements from the teams
James Drayson, CEO of Locai Labs, said the collaboration demonstrates how UK-developed foundational AI can help accelerate scientific research and industrial deployment. Pairing Locai's UK-built, energy-aware AI stack with First Light's fusion program aims to move the sector closer to clean energy outcomes.
Mark Thomas, CEO of First Light Fusion, noted the company's long history with ML for solution optimisation. As the fusion programme scales, he said AI will help deepen insight into the physics and speed up how the team tests ideas, refines hypotheses, and gets from results to implementation-reflecting British science and technology working together for a sustainable future.
Governance and next steps
The MoU establishes a framework for technical feasibility studies and forms a joint working group to oversee the initial phases. Findings from this trial will shape the scope of any longer-term partnership.
Practical checklist for labs considering similar pilots
- Define evaluation metrics upfront (e.g., simulation throughput, code quality, time-to-experimentation).
- Plan HPC integration: schedulers, storage, model placement, and resource allocation policies.
- Set data governance rules for on-prem inference, redaction, and provenance tracking.
- Establish human-in-the-loop guardrails for code suggestions and experiment orchestration.
- Invest in enablement: SOPs, usage patterns, and team training to convert pilots into practice. See the AI Learning Path for Research Scientists for structured upskilling.
Context
Inertial fusion and high energy density physics demand heavy simulation workloads and tight data controls. The UK's strategy for fusion research underscores the importance of industrial pathways and secure infrastructure; more on national objectives can be found in the UK government's fusion strategy here.
Your membership also unlocks:
