Fractile, a British startup building chips for artificial intelligence, has raised $720 million in a Series C. The round brings new backing from Britain’s Sovereign AI Fund and gives the company more resources as it works toward delivering its first chips and systems.
A major funding round for a hardware bet
The financing was led by Thrive Capital, Founders Fund, Lightspeed Venture Partners and Redpoint. Britain’s government said Fractile’s valuation was “over £5 billion.” The Financial Times and a Reuters source both reported a figure of $7 billion.
The government described its investment as a way of “ensuring the U.K. has a greater stake in the hardware that will underpin its future economy.” That framing places Fractile’s fundraising within a broader effort to build a stake in the technology infrastructure behind future AI use.
Fractile joins other well-funded startup efforts in purpose-built AI chips, alongside Axelera, Etched and OLIX. The field is attracting attention because specialized chips are one route companies are exploring to reduce the cost of inference: running increasingly demanding AI models in production.
Chips and software designed to work together
Fractile’s pitch extends beyond the chip itself. The company is pursuing a full-stack approach, pairing its hardware with a machine learning compiler intended to maximize the speed at which large language models, or LLMs, run on its chips.
That combination reflects the challenge of making AI systems useful in practice. Hardware provides computing capacity, while software can shape how effectively a model uses it. Fractile’s strategy is to develop both pieces together rather than focus on a chip alone.
The company presents itself as a frontier AI company, and its latest round was priced accordingly. That positioning sits alongside its more specific ambition: to build hardware that can handle the demands of running large models, with the compiler helping make those models execute faster.
Stacked DRAM is central to the design
Fractile’s main hardware distinction is its approach to DRAM, the most common type of RAM. Following a pivot, the company’s design now centers on stacking DRAM on top of each chip. Fractile hopes that this arrangement can get around the usual tradeoff between memory and capacity.
That tradeoff is becoming more pressing as AI systems take on longer chains of tasks. Agentic AI can require large models to work through extended sequences without sacrificing speed. When computing capacity struggles to keep pace, memory and performance become central considerations for the hardware running those models.
Purpose-built AI chips are one possible response to that strain, though the source notes that such chips have had limited availability so far. Fractile’s design is still a bet on a particular way to balance memory, capacity and speed; the funding gives the company room to move from that design toward shipping systems.
The next milestone is shipping
Fractile CEO Walter Goodwin acknowledged the distance between the company’s current position and a finished product. In a LinkedIn post, he wrote that “there is still a lot of work ahead.” He said the funding will enable the startup to “ship [its] first chips and systems next year.”
For Fractile, the round therefore marks both investor confidence and a step toward a practical test of its approach. Its stacked DRAM design and machine learning compiler will need to come together in chips and systems that can run demanding models at speed. The company’s stated next milestone is to ship those products next year.