英国 AI 芯片初创公司 Fractile 获 7.2 亿美元 C 轮融资
With $720M round, Britain’s Fractile gets a leg up in the race for AI chips
英国 AI 芯片初创公司 Fractile 完成 7.2 亿美元 C 轮融资,由 Thrive Capital、Founders Fund 等机构领投,英国主权 AI 基金参投,估值超过 50 亿英镑。Fractile 采用芯片堆叠 DRAM 的全栈架构并搭配自研编译器,旨在降低大语言模型推理成本。CEO Walter Goodwin 表示新资金将支持公司在明年交付首批芯片和系统。
原文披露了 DRAM 堆叠的定制芯片架构与交付时间线,展示了初创团队降低模型推理成本的硬件解法。

Fractile, a British startup developing AI chips, has raised a $720 million Series C backed by Britain’s Sovereign AI Fund with hopes of “ensuring the U.K. has a greater stake in the hardware that will underpin its future economy,” its government stated.
Alongside the likes of Axelera, Etched, and OLIX, this makes Fractile one of the top-funded startup players in the race for purpose-built AI chips. Such chips are one of several avenues being explored to reduce the cost of inference — aka using increasingly demanding AI models in production.
However, Fractile seeks to go beyond hardware with a full-stack approach that also includes a machine learning compiler to maximize the speed at which LLMs will get executed on its chips.
As a result, the company frames itself as a frontier AI company, and its latest round led by Thrive Capital, Founders Fund, Lightspeed Venture Partners, and Redpoint was priced accordingly. According to the British government, the valuation was “over £5 billion” — quite possibly $7 billion, a figure reported both by the Financial Times and a Reuters source.
Still, Fractile’s main differentiator will arguably be its hardware design. After a pivot, its approach now revolves on DRAM, the most common type of RAM, but stacked on top of each chip in such a way that it is hoping will bypass the usual trade-off between memory and capacity.
This trade-off has become more problematic with the rise of agentic AI, as companies seek to perform longer chains of tasks with large models without compromising on speed. As compute struggles to keep up, hopes are high for AI chips, despite their limited availability to date.
In a LinkedIn post, Fractile CEO Walter Goodwin wrote that “there is still a lot of work ahead” for the company, and that the new funding will enable the startup to “ship [its] first chips and systems next year.”
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
来源:TechCrunch AI · techcrunch.com