The UK is launching two major research labs backed by £60 million in funding to develop artificial intelligence systems that can power physical robots and other embodied AI without requiring enormous centralized computing power. The labs, led by University College London and the University of Oxford, represent a strategic shift in how Britain approaches AI development, focusing on making the technology more accessible, efficient, and independent from the handful of large technology companies that currently dominate the field.

Why Is the UK Betting on Embodied AI and Robots?

One of the two labs, called the British Open-ended Learning and Discovery (BOLD) Lab, will specifically focus on embodied systems like robots. This lab will fundamentally rethink how AI systems learn, moving away from the current approach where AI struggles with real-world complexity. Instead, BOLD will develop systems capable of working alongside humans, navigating physical environments, and operating without vast centralized computing power. This is a significant departure from today's AI, which typically requires enormous data centers and massive computational resources to function effectively.

The other lab, the Science of Fundamental AI Research (SOFAIR) Lab, will develop next-generation open-source AI technologies designed to run on widely available hardware. Modern AI currently relies heavily on a small number of popular architectures trained on vast amounts of data, requiring immense computing infrastructure. SOFAIR will explore new AI architectures and those specifically designed to run on hardware that researchers and institutions already have access to, making cutting-edge AI more widely accessible.

What Are the Three Research Pillars of the BOLD Lab?

The BOLD Lab, led by Associate Professor Jakob Foerster at the University of Oxford, will be built around three ambitious research areas that directly address the challenges facing physical AI development today:

• New Learning Algorithms: Developing fundamentally different ways for AI systems to learn, rather than simply scaling up existing methods with more data and computing power.
• Human-Centered AI: Creating systems that can work effectively alongside humans, understanding human needs and preferences in real-world contexts.
• Embodied Systems: Building robots and other physical AI systems that can navigate and operate in real environments without relying on remote data centers or constant cloud connectivity.

"The UK cannot win the global AI race simply by trying to outspend the largest technology companies on data and compute. BOLD is about a different route: discovering fundamentally new ways to build AI that are more efficient, more open and better aligned with human needs," said Associate Professor Jakob Foerster.

Associate Professor Jakob Foerster, University of Oxford

How Will These Labs Support Commercialization and Talent Development?

Both labs will actively support the commercialization of their research with targeted support for entrepreneurship and spin-outs, meaning the breakthroughs developed in these academic settings will have a pathway to becoming real products and services. The labs will also invest heavily in developing the next generation of AI researchers and engineers.

Each lab will receive around £8 million initially, with further funding released following an assessment in autumn 2026. This phased approach ensures early progress and robust plans are in place before the remaining investment is confirmed. Additionally, £2 million per lab is earmarked to support a minimum of ten doctoral students, with investment in researchers at every career stage, from doctoral training through to postdoctoral support and academic staff.

The labs will work alongside leading players in the UK AI ecosystem, including The Alan Turing Institute and UKRI's AI research hubs, building on existing centers of excellence and strong links with industry partners. This collaborative approach aims to translate fundamental advances into real-world impact across sectors including healthcare, education, small businesses, public services, science, and advanced industry.

What Makes This Investment Different From Other AI Research?

This announcement is particularly significant because it reflects a deliberate strategic choice by the UK to pursue a different path in AI development. Rather than trying to outspend large technology companies on raw computing power and data, these labs will focus on discovering fundamentally new approaches to AI that are more efficient and better suited to real-world applications like robotics.

"We are only just beginning to unlock AI's huge potential to grow our economy and improve our public services. With our world-leading universities and deep pool of AI expertise, Britain can set the agenda for what comes next. These new labs will lead the world in the fundamental work that is set to make AI cheaper, more practical and easier to adopt so more businesses and public services across the UK can benefit," stated Professor Charlotte Deane.

Professor Charlotte Deane, Senior Responsible Owner for the UK Research and Innovation AI Programme and Executive Chair of EPSRC

The announcement also goes further than originally planned. UKRI doubled the number of labs from one to two and increased total investment from £40 million to up to £60 million, reflecting the organization's recognition of the significant opportunity identified through this program. This expansion demonstrates the UK's commitment to establishing itself as a global leader in AI research, particularly in areas where the country has a genuine chance to lead, such as AI that works across different types of systems and data, and approaches that enable people and AI to work together more effectively.

By building this capability at home, backed by world-leading universities, the UK aims to strengthen its own expertise, reduce reliance on large foreign technology providers, and secure Britain's place at the forefront of AI development. For the robotics and embodied AI sector specifically, this investment signals that the future of physical AI may depend less on having the most powerful computers and more on having the smartest algorithms and the most efficient learning approaches.