- UK AI chip company Fractile exits stealth with $15M in Seed funding.
- Founded in 2022 by Walter Goodwin, Fractile is developing a new AI chip.
- The chip aims to be 100x faster and 10x cheaper than current hardware.
- Fractile’s technology focuses on in-memory compute to enhance AI model inference.
- The company targets a 20x improvement in power efficiency (TOPS/W).
- Funding round led by Kindred Capital, NATO Innovation Fund, Oxford Science Enterprises, with participation from Cocoa, Inovia Capital, and angel investors.
- Total funding raised to date stands at $17.5 million.
- Fractile plans to use funds to grow its team and advance commercial AI accelerator development.
Main AI News:
Fractile, a UK-based AI chip innovator, has emerged from stealth mode, announcing a significant $15 million in Seed funding. Founded in 2022 by Walter Goodwin, a 28-year-old AI PhD, Fractile is set to develop its inaugural AI chip, which promises to deliver up to 100 times the speed and 10 times the cost-efficiency of current models.
Current AI chips, designed primarily for training rather than inference—the process of applying live data to trained models—are a major bottleneck in AI performance. This limitation not only makes running AI models costly but also hinders performance and potential advancements. AI model developers face challenges in delivering distinctive innovations due to these constraints.
Fractile’s approach diverges from traditional methods. Instead of specializing in specific workloads, Fractile aims to fundamentally transform computational operations. This strategy involves creating new chip architectures from innovative building blocks and developing scalable systems. By focusing on in-memory compute, Fractile eliminates the need for transferring model parameters between chips, thus enhancing efficiency and performance.
The new chip is designed to execute 99.99 percent of the necessary operations for model inference, targeting performance enhancements of 100 times in speed and 10 times in cost reduction. Additionally, Fractile’s system is expected to achieve a 20-fold improvement in power efficiency, measured in Tera Operations Per Second per Watt (TOPS/W). This will enable greater user capacity and faster response times, which is crucial for applications like language models and other complex AI tasks.
The $15 million funding round was led by Kindred Capital, NATO Innovation Fund, and Oxford Science Enterprises, with contributions from Cocoa, Inovia Capital, and notable angel investors including Hermann Hauser, Stan Boland, and Amar Shah. To date, Fractile has secured a total of $17.5 million in funding.
Walter Goodwin, CEO and Founder, emphasized the critical need for hardware advancements in the AI industry. “The limitations of existing hardware are the biggest barriers to better performance and wider adoption,” he stated. “Fractile’s technology will not only enhance speed and cost-effectiveness but also unlock new possibilities for solving complex global challenges.”
John Cassidy from Kindred Capital praised Fractile’s strategic vision and its capability to anticipate future AI requirements. “The team’s expertise positions them to meet evolving AI needs, making us eager to support their journey.”
Stan Boland, an angel investor, highlighted the potential of Fractile’s technology and its promise to become a major player in the AI hardware space. “Walter’s commitment to market-driven innovation and building a top-tier team will position Fractile as a leading partner for AI model providers.”
Fractile, with a team composed of senior experts from NVIDIA, ARM, and Imagination, has already filed patents for its novel circuits and in-memory compute technology. The company plans to use the new funding to expand its team and advance the development of its first commercial AI accelerator.
Conclusion:
Fractile’s recent $15 million Seed funding round signifies a pivotal shift in AI hardware technology. By focusing on in-memory compute and innovative chip architectures, Fractile is set to address critical performance and cost challenges in AI inference. This breakthrough could significantly impact the AI market by reducing operational costs, enhancing performance, and accelerating advancements in AI capabilities. The emphasis on power efficiency and scalable systems positions Fractile as a key player in the evolving AI hardware landscape, likely attracting interest from major AI model providers and setting new standards for the industry.
