Artificial Intelligence
Aims
The Cambridge Open Zettascale Lab is advancing AI as a transformative tool across science, engineering, and industry. This theme focuses on the development, optimisation, and deployment of AI models on high-performance computing infrastructure—enabling breakthroughs in areas such as personalised medicine, climate modelling, materials science, and fusion energy.
Research under this theme includes scaling foundation models and generative AI architectures, accelerating training and inference on heterogeneous compute systems, and integrating AI with simulation and experimental workflows. The Lab also explores responsible AI, with attention to safety, interpretability, and energy-efficient computing.
Through collaborations with national and international partners, COZL is helping to build an AI research ecosystem that balances performance, openness, and accessibility—supporting the UK’s ambition to lead in trustworthy, compute-intensive AI.
The AI theme at the Cambridge Open Zettascale Lab is focused on advancing artificial intelligence at scale to accelerate scientific discovery and innovation. One of its key aims is to enable scalable training and inference of large AI models—such as foundation models and generative AI—by optimising performance across high-performance computing (HPC) infrastructure. This includes addressing challenges in parallelism, memory usage, and throughput to ensure efficient use of cutting-edge compute systems.
Another central goal is to integrate AI more deeply into scientific workflows. By combining machine learning with traditional simulations and experimental data, the Lab supports hybrid modelling approaches that can significantly enhance research in areas such as fusion energy, climate science, and personalised medicine. Benchmarking and evaluation of emerging AI hardware and software platforms is also a core activity, helping to inform future system design and procurement by comparing performance, efficiency, and usability across diverse architectures.
The Lab is committed to advancing responsible and sustainable AI. This includes exploring methods to reduce the environmental impact of large-scale AI, promoting interpretability, and ensuring AI development aligns with ethical standards and regulatory frameworks. In parallel, the Lab actively fosters open research and skills development by contributing to open-source tools, datasets, and models, while offering opportunities for hands-on collaboration, training, and mentorship.
Finally, the AI theme supports broader national infrastructure efforts such as the UK’s AI Research Resource (AIRR), contributing technical expertise and compute capability to ensure that academic and industry communities can access and benefit from sovereign, high-performance AI platforms.