Fractile secures £577M to develop SRAM-based AI chips, challenging Nvidia

UK-based AI chip startup Fractile has reportedly raised £577 million in a Series C round, according to VentureBurn, just months after closing a $220 million Series B. The funding is said to be aimed at expanding its engineering teams in London and Bristol and advancing its SRAM-based chip architecture. VentureBurn also reports that the company has signed a supply agreement with Anthropic, the AI lab behind Claude, with deliveries expected in 2027.

If accurate, Fractile’s approach would mark a departure from Nvidia’s GPU-focused model. The company is reportedly betting on static random-access memory (SRAM) to improve inference performance, arguing that it could reduce latency and energy consumption compared to traditional high-bandwidth memory solutions. VentureBurn suggests this design could offer efficiency gains, though such claims have yet to be demonstrated in large-scale deployments.

The size of the round—reportedly tripling the company’s previous raise in under six months—reflects strong investor interest in AI chip alternatives. This trend aligns with recent funding rounds for other Nvidia challengers, including FuriosaAI, which raised $620 million last month at a $2.9 billion valuation, as StartupReader previously reported. While FuriosaAI’s focus appears to be on data center power efficiency, Fractile’s reported SRAM-based design targets memory access speed, indicating that investors are exploring multiple technical paths to compete with Nvidia.

Anthropic’s reported involvement could signal demand for non-Nvidia hardware among major AI labs. If the supply agreement holds, it may provide Fractile with a high-profile customer and revenue path, though such partnerships are rare for early-stage deep-tech startups. For Anthropic, the deal could offer a potential alternative to Nvidia’s ecosystem, though the practical benefits remain to be seen.

Nvidia’s dominance in AI chips has created opportunities for challengers, as customers face supply constraints and high costs. StartupReader’s coverage of Nvidia’s recent GTC Berlin event noted the company’s shift toward framing AI as a deployment challenge rather than just a hardware one. Fractile’s reported focus on inference—rather than training—could position it in a segment where Nvidia’s lead is less entrenched, though Nvidia’s software and ecosystem advantages remain significant barriers.

Fractile’s reported raise also highlights growing investor confidence in European deep-tech startups. If successful, it could signal a broader shift away from single-supplier dependency in AI hardware. This trend has been echoed in recent funding rounds for other startups, such as SiMa.ai and Volantis, which target niche segments of the AI chip market.

What to watch next: Whether Fractile can deliver on its reported SRAM-based architecture and meet its 2027 timeline, and how Nvidia might respond to emerging competition. The startup’s next moves—whether another funding round, commercial partnerships, or potential acquisition—could further shape the AI chip landscape. For now, the reported raise suggests that the race for AI hardware dominance is far from settled.

Sources: ventureburn.com