British AI inference chip startup Fractile has secured $220 million in Series B funding. The round was co-led by Accel, Factorial Funds, and Founders Fund, with participation from Conviction, Gigascale, O1A, Felicis, Buckley Ventures, and 8VC.
Founded in 2022 by Oxford-trained chip engineer Walter Goodwin, Fractile focuses on inference-the time it takes for trained AI models to generate outputs. As models scale, they require millions of tokens, which increases latency due to data movement between processor and memory.
Fractile’s novel logic chip attaches memory within a standard server rack to reduce latency and maximize bandwidth without sacrificing speed. Goodwin told the Wall Street Journal the design avoids traditional high-bandwidth memory and on-chip SRAM, suggesting an entirely new architecture.
The chips aim to enable workloads impossible on standard GPUs, compressing months of work into days or lab computations into coffee breaks. Goodwin wrote that the 21st century will be defined by inference driving breakthroughs in drug discovery, software engineering, and materials science.
Fractile enters a competitive market dominated by Nvidia, with rivals including Cerebras (set for a $5.5 billion IPO), SambaNova, Untether AI, Graphcore, and cloud giants AWS and Google Cloud. OpenAI is also developing its own inference chip with Broadcom and TSMC.