Fractile, a London- and Bristol-based startup building chips designed specifically to run already-trained AI models, said this week it has raised $720 million in a Series C round at a $7 billion valuation — a more than sixfold jump from the roughly $1.1 billion the company was worth after its last round just six months ago.
The round was led by Thrive Capital, Founders Fund, Lightspeed Venture Partners and Redpoint, with Alphabet's independent venture arm CapitalG, BlackRock, DST Global, the U.K.'s own sovereign AI fund and the British Business Bank also participating, according to a Reuters report cited by SiliconANGLE. Fractile had previously raised a $220 million Series B in May at roughly $1.1 billion.
Founded in 2022 by Walter Goodwin, an Oxford-trained AI researcher, Fractile is chasing a narrower problem than general-purpose chipmakers like Nvidia: inference, the stage after a model has already been trained, when it's actually answering a user's question. Its pitch is architectural — rather than pairing a processor with high-bandwidth memory sitting beside it on the same package, Fractile stacks ordinary DRAM directly on top of the logic itself, shortening the distance data has to travel. The company says that approach could let large language models process several thousand tokens per second per chip, with up to 100 times today's performance and a 90% cut in energy use, though those figures are the company's own projections rather than independently verified benchmarks.
Fractile has not shipped a production chip. It expects to begin manufacturing its first commercial parts in 2027, selling them bundled into rack-scale systems with its own software rather than as standalone components — a go-to-market plan similar to rivals including Axelera, Etched and Olix, all of which are also racing to challenge Nvidia's hold on AI hardware. CryptoBriefing reported that Anthropic has shown early interest in Fractile's chips, though the outlet cautioned no deal has been finalized.
The U.K. government, an investor in the round through its sovereign AI fund, framed the deal in national-security terms. Its statement said Britain should be in a position where it "has a greater stake in the hardware that will underpin its future economy" — part of a broader push by London to avoid depending entirely on American and Taiwanese chip supply chains as AI computing demand accelerates.
