AMD pledges up to £2 billion for UK AI supercomputers, research partnerships

Advanced Micro Devices is committing up to £2 billion over the next five years to expand artificial intelligence research and compute resources in the United Kingdom, the company announced Monday. AMD said the program is meant to accelerate AI innovation, broaden access to advanced computing, and support scientific work tied to areas such as healthcare and climate modeling. AMD made the announcement at London Tech Week, positioning the pledge as part of broader UK efforts to build “sovereign” AI infrastructure and talent.

AMD’s plan includes new university and research collaborations, anchored by a partnership with Imperial College London. In the company’s release, AMD said it will work with the university to advance computational science and research that depends on large-scale computing resources, and it plans to explore ways to optimize AI models, scientific workflows, and data-intensive applications on AMD compute platforms. AMD also pointed to ROCm, its open software stack used to run and optimize GPU-accelerated workloads, as part of the effort.

A separate track pairs AMD with Oriole Networks to support the UK’s ARIA Scaling Inference Lab, an initiative aimed at addressing practical bottlenecks in running AI models at scale. AMD said the work uses Oriole’s PRISM photonic networking architecture together with AMD Instinct GPUs and AMD EPYC processors to evaluate approaches for scaling inference workloads, with targets that include performance and energy efficiency and reduced latency. AMD also described the effort as a step toward what it expects could become a large-scale AI system powered by a “pure photonic network.”

On national supercomputing capacity, AMD said it and Dell Technologies are working with the University of Cambridge on an expanding AI footprint that includes two systems: Zenith and Sunrise. Zenith is described as a new “AI-for-science” platform funded by the Department for Science, Innovation and Technology and UK Research and Innovation, designed and operated by Cambridge and built with AMD and Dell technology. Sunrise, AMD said, is a second AI supercomputer powered by AMD and Dell that is being built now, funded by the Department for Energy Security and Net Zero, owned by the UK Atomic Energy Authority, and operated by Cambridge.

AMD said the systems and research programs would draw on its Instinct GPU line for AI acceleration, its EPYC CPU line for general compute and data management, and ROCm for software support. The company said the initiatives are intended to span a broad set of use cases, including healthcare research, climate modeling, materials science, engineering simulation, and fusion research, as well as scientific AI model development.

AMD framed the investment as aligned with UK policy priorities including the AI Opportunities Action Plan and AI Hardware Strategy. In remarks included in the release, AMD Chair and CEO Dr. Lisa Su said the UK has the talent and ambition to lead the next era of AI, and that AMD wants to deepen its commitment by expanding access to compute infrastructure needed for “sovereign AI.” UK government leaders also welcomed the pledge, with Chancellor of the Exchequer Rachel Reeves calling it a vote of confidence, and Technology Secretary Liz Kendall linking it to compute capability, job creation, and advanced AI technology development.

AMD also said it is building on recently announced work with Oxford Quantum Circuits and JPMorganChase, citing a June 3 research collaboration to develop a quantum-AI platform in London. In that separate announcement, AMD said OQC’s GENESIS quantum system would be integrated with AMD-supported AI and classical compute, with JPMorganChase researchers testing near-term quantum and hybrid quantum-classical applications in a secure environment. For the current pledge, AMD did not say how much of the broader UK investment would be tied to quantum-related work or how soon the photonic networking and supercomputing systems will become available to new users. The company also did not disclose a breakdown of the £2 billion by project, hardware specifications, or procurement terms.

Looking ahead, the next benchmarks will likely come from how quickly AMD-supported systems move from construction to operational use. Observers may want to watch for updates on Zenith’s ramp-up and Sunrise’s build timeline, details on access schedules for researchers and public-sector partners, and whether the Scaling Inference Lab demonstrates measurable gains from photonic networking approaches. AMD’s role will also be tested by how effectively ROCm-optimized workflows perform on the deployed infrastructure across the kinds of research workloads the company highlighted.