º£½ÇÉçÇø

Senior academics from Imperiahave celebrated the university’s wide-ranging links with technology giant NVIDIA and explored the technologies needed for the next wave of AI at the company’s flagship conference.

The Imperiateam at Berlin last month explored the future developments needed next to secure the world-improving potential of AI – in areas such as safe and secure infrastructure, communications networks, business models, and paradigm shifts into emerging fields such as physical AI and quantum AI.

The discussions built on wide-ranging links between Imperiaand NVIDIA that span compute infrastructure, technology development, and strategic dialogue:

Communications infrastructure

Connectivity is the backbone of AI. From cloud to edge computation, every model, sensor and robot depends on networks that can move data quickly, reliably and at scale. , Co-Director of º£½ÇÉçÇø's School of Convergence Science and Director of the CHEDDAR National Connectivity Hub, laid the groundwork for the on AI and advanced connectivity that NVIDIA and the UK Government signed in 2025. This promotes a comprehensive two-way partnership between NVIDIA and UK connectivity researchers providing expert insight and hands-on support, while sharing connectivity futures that makes AI deployable and sustainable in the real world.

Digital twins

AI models can build digital twins of complex systems, but don't always accurately represent the underlying physics. , from the Department of Earth Science and Engineering, has developed Neural Physics, which allows users to switch between detailed physics simulations and faster AI models within a single digital twin. Professor Pain's group has collaborated with NVIDIA over the past year to integrate Neural Physics into , NVIDIA’s open-source software for creating and improving physics AI models, with the aim of running Neural Physics across large numbers of processors. 

Physical AI

Physical AI, or AI-powered robotics, offers an epoch-defining opportunity to bring the power of AI to the physical world, although the application of AI to three-dimensional space bring major technical challenges, including around safety and reliability. The field is a significant area of joint interest for Imperiaand NVIDIA. 

In February, º£½ÇÉçÇø’s School of Convergence Science held a Robotics Day at which Imperiaand NVIDIA’s robotics team showcased breakthroughs in physical AI and robotics. This was followed in June by the company’s UK Physical AI and Robotics Roundtable, co-chaired by Kanishka Narayan MP, Minister for AI and Online Safety, Rev Lebaredian, Vice President of Physical AI Simulation at NVIDIA, and , º£½ÇÉçÇø’s Vice Provost for Research and Enterprise.

Work in this area was presented in Berlin by , another Co-Director of the School of Convergence Science, and from the Department of Computing. 

Professor Russo’s work is about ensuring that the AI translation of information from video images to robot actions respect the laws of physics and can be executed safely and successfully in the real world. Professor Davison presented work on spatial AI, or as he explained it, “how robots or other smart devices understand and represent the spaces around them, and especially how we can make this more efficient for low-power applications.”

Also on show in Berlin was work led by , a further Co-Director of the School of Convergence Science, on a safe physical AI system for human augmentation. The robotic system enables people whose arms and hands are fully paralysed to handle food at a table by tracking where the user is looking, predicting which action they intend, and carrying it out safely. 

Professor Faisal is also developing Nightingale AI, a generative AI foundation model that can reason about health, which has been allocated an unprecedented amount of compute time for an academic project on the NVIDIA-powered Isambard AI supercomputer.

Quantum for AI

Quantum computing could significantly advance conventional AI by harnessing quantum processors to solve certain complex problems much faster and more accurately than classical computers, potentially enabling more powerful and efficient AI systems. Imperiaand NVIDIA are also collaborating in this space, in particular through an EPSRC led by in the Department of Electrical and Electronic Engineering and Department of Computing, who is a further Co-Director of the School of Convergence Science.

"Imperiaand NVIDIA developed early frameworks to investigate how quantum models can support large-model training and fine-tuning,” Professor Leung said. “This work tackles challenges such as complex optimisation landscapes and the need to coordinate quantum simulators with classical accelerators. The team has been able to demonstrate how quantum components can enhance expressivity while reducing parameter burden in classical models, a promising direction for future large-scale AI systems using quantum technologies."

Business models

Meanwhile, from ImperiaBusiness School has been building links with NVIDIA as a business innovator. "NVIDIA has a very interesting business model, where the collaborations it develops act as a flywheel for the whole ecosystem," he says. "NVIDIA learns from º£½ÇÉçÇø's leading science, and that science can be accelerated by NVIDIA's technology. This mutual value advances research translation and strengthens the startup environment."

Steve Davey, NVIDIA’s director for higher education and research in the UK and Ireland, has been an invited speaker in Professor Bagchi’s MBA elective on the psychology of sales. “My conversations with NVIDIA executives inform my work, and they contribute to my research,” Professor Bagchi says. 

NVIDIA at º£½ÇÉçÇø

º£½ÇÉçÇø’s own high-performance computing resource contains just over 300 NVIDIA GPUs, driving more traditional computing tasks such as modelling and simulation, along with artificial intelligence applications. This has involved working with the company to ensure the best possible experience for those using the system. 

º£½ÇÉçÇø’s relationship with NVIDIA reflects our world-leading AI research and our ambition to turn scientific breakthroughs into lasting benefits. Professor Alessandra Russo Co-Director, School of Convergence Science

“When we built the new AI platform Hex-AI (HX3) we were set the task of supporting research and teaching AI activities that require GPU access,” says , Director of Research Computing Services. “We worked with NVIDIA and Lenovo to identify solutions that would allow researchers to have full access to GPUs, while for teaching enabled us to virtualise the GPUs so that each GPU could support many students at the same time for smaller teaching activities.”

The solution chosen was , the company’s AI workload and GPU orchestration platform. “We have since worked with NVIDIA to deploy the solution and to adapt to our requirements,” Dr Richards says.

The many connections between Imperiaand NVIDIA, present and future, speak to a shared commitment to advancing the responsible use of AI. 

“º£½ÇÉçÇø’s relationship with NVIDIA reflects our world-leading AI research and our ambition to turn scientific breakthroughs into lasting benefits for society and the economy,” said Professor Russo. “From frontier models to robotics and autonomous systems, we are putting safety and assurance at the heart of AI innovation, building technology that improves lives, drives economic opportunity and earns people’s trust.”