The company plans to use the funds to speed up the development and release of its first chips.
UK-based AI inference chip startup Fractile has secured $220m in a Series B funding round to advance its efforts in developing next-generation inference hardware for AI systems.
The funding round was led by Accel, Founders Fund, and Factorial Funds, with additional investment from Gigascale, Conviction, Buckley Ventures, O1A, 8VC, Felicis, and existing backers.
Access deeper industry intelligence
Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.
Fractile plans to use the funds to speed up the development and release of its first chips and systems to customers, as well as to expand its team in the UK, US, and Taiwan.
UK AI Minister Kanishka Narayan said: “This investment is a real milestone for Fractile and a strong vote of confidence in British AI. It shows that UK companies at the cutting edge are pulling in global investment while anchoring high value jobs and expertise here at home – powering a tech sector already worth $1tn.”
In February 2026, Fractile announced plans to invest £100m in its UK operations over a three-year period. Funds have been allocated to expand its sites in London and Bristol and to establish a new hardware engineering centre in Bristol.
The investment aligns with government efforts to boost the UK’s AI sector, including moves to position the country as a global centre for AI and data infrastructure.
Founded in 2022, Fractile aims to address the speed limitations faced by advanced AI models when generating usable outputs.
The company is focused on rethinking hardware architectures to enhance the speed and economic viability of inference tasks, a core limitation for today’s AI systems.
According to Fractile, the bottlenecks in memory bandwidth and inference speed have become significant obstacles as large language models now generate outputs with as many as 100 million tokens.
On current hardware, such outputs can require up to a month to process, but Fractile says new approaches are needed to achieve greater throughput and reduce this time to a single day.
Fractile operates across the AI stack, from foundational research through to hardware design and process innovation, developing systems intended to improve both cost and latency of inference.