Fractile raises $720 million ahead of planned first AI chip shipments in 2027
The round values the British startup at more than £5 billion. Its pitch combines a different memory design with software built to run large language models faster.
Loading page…
The round values the British startup at more than £5 billion. Its pitch combines a different memory design with software built to run large language models faster.
Listen to this story
Fractile is pairing a chip architecture that stacks DRAM on each processor with a machine-learning compiler, aiming to improve inference speed without the usual capacity-versus-bandwidth tradeoff. The London-and-Bristol startup’s $720 million Series C—reported by the UK as more than £500 million—values it above £5 billion and backs development, not a product already available. First chips and complete systems remain a company target for 2027.
Thrive Capital, Founders Fund, Lightspeed Venture Partners and Redpoint led the round, according to TechCrunch.
The UK Sovereign AI Venture Fund and British Business Bank participated, but the size of either public investment was not disclosed.
The British Business Bank says its capital is intended to help scale Fractile’s hardware in the UK; Business Secretary Jonathan Reynolds cited planned Bristol expansion.
British AI chip startup Fractile has raised $720 million to help bring its first hardware to market. The Series C, announced November 11, includes British government backing and values the company at more than £5 billion. CEO Walter Goodwin says the money will enable its first chips and systems to ship in 2027.
Fractile, based in London and Bristol, is developing hardware for AI inference: running trained models to produce responses. Purpose-built chips are one route companies are pursuing to reduce that cost. Fractile’s approach also includes software, rather than relying on a new processor alone.
After changing its hardware approach, the company settled on a design that stacks DRAM, a common type of computer memory, on top of each chip, TechCrunch reported. Fractile hopes that arrangement will overcome the usual tradeoff between memory bandwidth—the speed of moving data—and the amount of memory available.
That tradeoff becomes more pressing as companies ask AI agents to carry out longer chains of tasks with large models while maintaining speed. Fractile is also developing a machine-learning compiler, software intended to make large language models execute faster on its chips. The hardware and compiler are parts of the same development effort.
Thrive Capital, Founders Fund, Lightspeed Venture Partners and Redpoint led the Series C, according to TechCrunch. The British government separately confirmed participation by its Sovereign AI Venture Fund and the British Business Bank, alongside investors including CapitalG, Lightspeed and Redpoint.
The government’s announcement described the round as worth more than £500 million and put Fractile’s valuation at more than £5 billion. Those are figures for the overall financing and company value, not the size of the public investment. The announcement did not disclose how much either state-backed investor contributed.
For the government, the investment serves an industrial goal as well as a technology ambition. It says backing domestic chip designers can reduce dependence on overseas technology and supply chains, while keeping more skilled jobs, expertise and economic returns in Britain. Those are the intended benefits of its support, rather than outcomes already delivered by Fractile.
For too long, Britain has produced some of the world’s best scientific ideas, only to see too much of the technology, industry and wealth they create built elsewhere. We are determined to change that.
Kanishka Narayan, UK AI minister
The British Business Bank describes its role as providing the long-term capital Fractile needs to scale its hardware technology in the UK. Business Secretary Jonathan Reynolds also pointed to the company’s plans to expand its presence in Bristol. That links the financing to an existing British engineering base, not just a future market for the chips.
Fractile’s stated scope stretches from chips through software to fundamental machine-learning research. The government presents that combined approach as a way to make AI hardware faster, cheaper and more energy-efficient. The financing backs development across those layers; it is not an announcement that a finished processor is now available to customers.
Goodwin acknowledged the remaining development work in a LinkedIn post quoted by TechCrunch, writing that “there is still a lot of work ahead.” His next milestone is shipment of both chips and systems, rather than a chip design alone. The schedule remains the company’s plan, with delivery still ahead of it.
Loading discussion...
Join the conversation
Explain whether domestic capability justifies taking that early investment risk.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.