Test report DSG-8593 · Rev D · tested November 12, 2026

Processors & AcceleratorsDevice under test

Fractile valued at $7bn after $720m Series C round

Fractile raised $720m at a $7bn valuation from Thrive, Founders Fund and UK state funds, and plans to ship its first in-memory compute chips next year.

Read
2 min
Words
487
Node
28nm
Operator
Amara Osei

Spec summary

  1. Fractile raised $720m in a Series C at a $7bn valuation
  2. Thrive Capital, Founders Fund, Lightspeed and Redpoint led the round
  3. Founded in 2022 by Oxford PhD student Dr. Walter Goodwin
  4. Company claims 20x higher speeds from DRAM memory
  5. £100m ($135m) UK expansion announced nine months ago; first chips ship next year
UK chip startup Fractile valued at $7bn following $720m Series C funding round
Fig. AUK chip startup Fractile valued at $7bn following $720m Series C funding round — StartupStockPhotos / pixabay

UK chip startup Fractile has closed a $720 million Series C round at a $7 billion valuation, with first silicon and systems scheduled to ship next year.

Thrive Capital, Founders Fund, Lightspeed and Redpoint led the round. Dragoneer, DST Global, BlackRock, T. Rowe Price and CapitalG joined as new investors. Accel, Conviction and Felicis returned.

Two public-sector backers also participated: the UK government's Sovereign AI venture fund and the British Business Bank. Neither disclosed the size of its investment.

Who is Fractile and what does it build?

Dr. Walter Goodwin founded the company in 2022 while a PhD student at the University of Oxford's Robotics Institute. Fractile develops chips based on in-memory compute, an architecture that runs calculations directly in computer memory rather than shuttling data between memory and processor.

Speaking to DCD in 2024, Goodwin said the approach aims to cut power consumption, raise performance, and deliver faster and cheaper inference at scale — the workload that dominates large AI model deployment.

The company claims one measurable result so far: extracting 20x higher speeds from DRAM memory, which Goodwin calls key to running models much faster than today's chips allow.

What did Goodwin say about the round?

"I have been so lucky to get to build and lead an extraordinary team at Fractile, who span the breadth of what is needed to pursue this mission, from silicon to software to systems," Goodwin wrote on LinkedIn.

He paired that praise with a caution about the distance still to travel: "While we've already scored some key breakthroughs — like our work to extract 20x higher speeds from DRAM memory, which is key to running models much faster than today's chips can — we have very big ambitions, and there is still a lot of work ahead. This support will see us ship our first chips and systems next year."

That shipping date gives the $7 billion valuation a concrete test: the company has not yet delivered commercial hardware.

What is the UK expansion behind the deal?

The state backing arrives nine months after Fractile announced plans to invest £100 million ($135m) in its UK operations over three years. The plan includes:

  • Growth of the existing London site
  • Growth of the existing Bristol site
  • A new hardware engineering facility in Bristol

The Sovereign AI fund and British Business Bank participation ties the round to UK industrial policy, placing public money alongside US venture firms at a $7 billion private mark.

Founded three years ago by a then-PhD student, Fractile now carries a valuation that places it among the better-funded semiconductor startups pursuing inference-specific silicon. The in-memory compute route it has chosen targets the two costs that define large-model deployment: power and memory bandwidth.

First chips ship next year; until then, the $7 billion figure rests on a 20x DRAM speed claim and a team that spans silicon, software and systems.

via datacenterdynamics.com (Source)

Filed under

  • fractile
  • in-memory-computing
  • ai-inference
  • semiconductor-funding
  • dram
Share this article:

More from Amara Osei

Amara Osei

Show full bio

Staff writer covering money, funding and market shifts around hardware layer of artificial intelligence at Die Signal.

593 articles

Same lot · LOT-3628

« Previous articleNext article »