UK startup Fractile raises $720 million to develop AI chips

Chips and software

Fractile is one of the best-funded startups developing specialized AI chips. Axelera, Etched and OLIX are also working in this area. These chips are one way to reduce the cost of inference: using increasingly demanding AI models in production systems.

The company is developing not only hardware but also a machine-learning compiler designed to speed up the execution of large language models on its chips. Fractile describes this approach as developing advanced AI technology.

A key difference in the startup’s hardware architecture is its use of DRAM, a common type of random-access memory. After changing direction, the company began placing memory directly above each chip. Fractile expects this to overcome the usual trade-off between memory capacity and bandwidth.

The problem is especially apparent as agentic AI develops: companies need large models to run long sequences of tasks without losing speed. Computing resources remain scarce, and specialized chips are still available only in limited quantities.

Fractile CEO Walter Goodwin said the company still has a lot of work ahead. The new funding, he wrote on LinkedIn, will allow the startup to ship its first chips and systems next year.

Dataist's take: The round gives Fractile the resources to move from developing its architecture to shipping its first systems, but it does not by itself solve the problem of AI chip availability. The practical value of its approach will become clearer after the deliveries planned for next year.