SOFAIR Lab
EPSRC-funded AI research lab at University College London, tasked with developing open-source AI systems that run on commodity hardware, reducing dependence on frontier chips and major model providers.
Last refreshed: 24 June 2026 · Appears in 1 active topic
Can SOFAIR's commodity-hardware approach compete with frontier AI labs on major benchmarks?
Timeline for SOFAIR Lab
Launched to develop open-source AI running on commodity hardware
UK Startups and Innovation: EPSRC doubles AI-lab spend to £60mBackground
SOFAIR Lab was launched on 23 June 2026 as one of two new EPSRC-funded AI research labs announced as part of a £60m national AI research programme, doubled from a previous £40m commitment. The lab's mandate is to develop open-source AI systems capable of running on commodity hardware, reducing the academic and commercial research community's dependence on a handful of frontier model providers.
SOFAIR Lab is led by University College London and received approximately £8m in initial funding, alongside £2m for at least ten doctoral students. The programme is subject to an autumn 2026 review before further tranches are released. AI Minister Kanishka Narayan announced the launch.
The lab's open-source, commodity-hardware focus places it at the opposite end of the compute spectrum from frontier labs building on dedicated clusters of high-end GPUs. If successful, this approach could lower the barriers to AI research for institutions that cannot access proprietary supercomputing infrastructure.
What is the SOFAIR Lab and what is it trying to do?
How much funding did EPSRC give to the new AI labs?
Why does the UK want AI that runs on commodity hardware?
Written with AI from Lowdown's published reporting. Not individually reviewed by an editor. Updated 24 June 2026. Editorial standards.