UK AI chip startup Fractile raises $720M Series C backed by Sovereign AI Fund

Fractile, a British startup building chips to run large AI models, has raised a $720 million Series C with backing from the UK's Sovereign AI Fund. According to TechCrunch, the round was led by Thrive Capital, Founders Fund, Lightspeed Venture Partners and Redpoint.

The British government said the round values Fractile at "over £5 billion," and TechCrunch reports that both the Financial Times and a Reuters source put the figure at roughly $7 billion.

The government framed the investment as a way of securing the UK a greater stake in the hardware expected to underpin its future economy — a mark of how directly state capital is now flowing into AI infrastructure.

A full-stack play

Fractile is not betting on silicon alone. According to TechCrunch, the company is also developing a machine learning compiler intended to maximize how fast large language models execute on its chips, and it positions itself as a frontier AI company rather than a component maker — a framing that TechCrunch says was reflected in how the round was priced.

Stacked DRAM at the core

The hardware design remains the company's most notable differentiator. After an earlier pivot, Fractile's approach centers on DRAM — the most common type of RAM — stacked on top of each chip in an arrangement the company hopes will bypass the usual trade-off between memory and capacity.

That trade-off has grown more acute with the rise of agentic AI, as companies look to run long chains of tasks on large models without giving up speed. General-purpose compute has struggled to keep pace, and purpose-built AI chips are one of several avenues being explored to reduce the cost of inference — using trained models in production — even though such chips have so far been available only in limited volumes.

A crowded race

The round places Fractile among the best-funded startups pursuing purpose-built AI chips, alongside Axelera, Etched and OLIX, according to TechCrunch. All are targeting the same underlying pressure point: the expense of running increasingly demanding models at scale.

First chips due next year

Writing on LinkedIn, Fractile CEO Walter Goodwin said the new funding will allow the startup to ship its first chips and systems next year, while cautioning that "there is still a lot of work ahead."

Why it matters

Inference cost is becoming the binding constraint on AI deployment: as models and agent workflows grow larger, the economics of running them in production matter as much as training them. Startups like Fractile are attacking that cost at the silicon level, and a design that eases the memory-capacity trade-off would be significant if it works in practice.

The involvement of the Sovereign AI Fund also signals a shift in how governments engage with AI — not just through regulation or procurement, but through direct equity stakes in strategic hardware companies. For the UK, the deal is a bid for relevance in a chip market otherwise dominated by US and Asian players, though with no shipped silicon yet, a multibillion-pound valuation still carries considerable execution risk.