Fractile Secures $15M Funding for AI In-Memory Computing

Chip startup Fractile has recently secured a significant funding round of $15 million for its innovative in-memory AI compute technology. The company boasts that its solution can execute the latest AI models over 100 times faster and at a fraction of the cost compared to existing hardware options.

The funding was a collaborative effort, with Kindred Capital, NATO Innovation Fund, and Oxford Science Enterprises leading the round. Notable angel investors, including Herman Hauser of ASM, Stan Boland of Icera and Five.ai, and Amar Shah of Wayve, also contributed to the investment.

Established in 2022 in London, UK, Fractile had previously raised $2.5 million before emerging from stealth mode. The startup has been strategically recruiting talent from industry giants such as ARM, Imagination Technologies, and Nvidia to bolster its team.

Fractile has taken steps to protect its technology through patent filings, safeguarding its unique approach to in-memory compute. The company is currently in talks with potential partners and anticipates securing partnerships prior to the production of its first commercial AI accelerator hardware.

Dr. Walter Goodwin, CEO and Founder of Fractile, emphasized the significance of their technology in the AI landscape, stating, "In today’s AI race, the limitations of existing hardware represent the biggest barrier to better performance, reduced cost, and wider adoption. Changing the performance point for inference allows us to explore new ways to leverage AI models for solving complex problems."

Angel investor Stan Boland expressed confidence in Fractile's potential, highlighting the company's commitment to building a top-tier team of experts across semiconductor, software, and tools domains. Boland believes that Fractile is poised to become a trusted partner for major AI model providers in the near future.

Fractile distinguishes itself by focusing on in-memory compute, aiming to achieve 20 times the efficiency of any current data center AI inference system. This approach enables more simultaneous users to be served per inference system, leading to higher throughput and improved cost-effectiveness.

The company's technology opens up possibilities for advanced AI applications, such as recursive queries, chain of thought prompting, and tree search, which can enhance the quality of answers generated by AI systems.

Moreover, the enhanced performance offered by Fractile's solution has the potential to accelerate progress in various fields, including drug discovery, climate modeling, and video generation, by enabling faster and more efficient AI computations.

While Fractile is making waves in the AI hardware space, other companies like Axelera, Femtosense, and GraphCore are also exploring in-memory compute architectures for embedded and edge AI applications, indicating a broader industry trend towards more efficient and powerful AI hardware solutions.