The London and Bristol-based inference chip developer raised $720 million, months after a round that valued it near $1.1 billion
Fractile, a British startup building chips for artificial intelligence, has been valued at $7 billion in a fresh funding round. The British government is among the backers.
The company raised $720 million, according to a person involved in the deal who spoke to Reuters on Wednesday, Nov 11. That valuation places Fractile among the most valuable technology startups in Britain.
The details: big money, fast
The round is a Series C, and it lands a $7 billion post-money valuation on a company founded only in 2022. Walter Goodwin started Fractile, which splits its operations between London and Bristol.
In May 2026, Fractile closed a $220 million Series B at a valuation of approximately $1-1.1 billion. Roughly six months later, the price tag has climbed more than sixfold.
The investor lineup reads like a roll call of Silicon Valley heavyweights. Thrive Capital, Founders Fund, Lightspeed and Redpoint led the round, with CapitalG and BlackRock among the other participants.
The British state also wrote in. Government support came through the Sovereign AI Venture Fund and the British Business Bank, in a round that exceeded £500 million.
What Fractile actually builds
Fractile designs chips for AI inference. Inference is the step where a trained model actually does its job: answering a question, writing code, summarizing a document.
The company says its approach can run that workload faster while trimming both energy consumption and costs. Its design uses what it calls a memory-first architecture, placing memory closer to the processing unit.
Fractile has pitched its systems as capable of up to 90% lower operational costs for running large language models.
Fractile plans to begin data center deployments in 2027 and, as of November 2026, had not shipped any production chips.
The company has reportedly attracted interest from potential early customers, including Anthropic.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.
