Fractile Raises $720M Series C to Challenge AI Inference Chip Leaders

British AI chip startup Fractile has raised a $720 million Series C round, with Britain's Sovereign AI Fund among the backers, according to the British government. The valuation exceeded £5 billion — a figure the government stated publicly, and which the Financial Times and a Reuters source each placed at roughly $7 billion.

Thrive Capital led the round. Founders Fund, Lightspeed Venture Partners and Redpoint also participated. The government framed the investment as a step toward "ensuring the U.K. has a greater stake in the hardware that will underpin its future economy."

The round places Fractile among the best-funded startups building purpose-built AI chips, alongside Axelera, Etched and OLIX. These companies target one of the most expensive problems in modern computing: the cost of inference, the process of running trained AI models in production as the models grow more demanding.

What sets Fractile's hardware apart?

Fractile's core differentiator is a memory-centric chip design. After a strategic pivot, the company now builds its approach around DRAM, the most common type of RAM, stacked on top of each chip in a configuration intended to bypass the usual tradeoff between memory capacity and compute speed.

That tradeoff has become more pressing with the rise of agentic AI. Companies want to run long chains of tasks with large models without sacrificing speed, and conventional architectures struggle to keep memory bandwidth in step with compute demand. This is why investor interest in purpose-built AI silicon remains high despite the limited availability of such chips to date.

The stacked-DRAM approach puts Fractile in direct conversation with a broader industry effort to move memory closer to compute. The startup bets that collapsing the distance between the two will deliver faster large language model execution than general-purpose accelerators can offer.

Why does Fractile call itself a frontier AI company?

Hardware alone is not the full product. Fractile pairs its chips with a machine learning compiler designed to maximize the speed at which large language models execute on its silicon. This full-stack approach — chip plus compiler — is why the company positions itself as a frontier AI company rather than a component supplier.

That framing carried into the round's pricing. The investors that led the deal — Thrive, Founders Fund, Lightspeed and Redpoint — typically back frontier AI laboratories at multibillion-dollar valuations. Fractile secured a valuation normally reserved for model developers, on the strength of a silicon-plus-software stack that does not yet have shipping product.

When will the first chips arrive?

Production silicon is close but not here. CEO Walter Goodwin wrote in a LinkedIn post that "there is still a lot of work ahead" for the company. He said the new funding will let Fractile "ship [its] first chips and systems next year."

That timeline puts a hard marker on the company's next test. The Series C gives Fractile the capital to move from design to delivery, but the stacked-DRAM architecture must prove itself in working systems against incumbents and rival startups that already have hardware in customer hands.

The Sovereign AI Fund's participation also signals a second stakes race. Britain is buying a position in the AI hardware layer of its own future economy, and Fractile is now its most visible instrument in that effort. Whether the bet pays off will become clear when the first Fractile systems reach customers next year.