This funding round was led by Kindred Capital, NATO Innovation Fund, and Oxford Science Enterprises, with additional contributions from Cocoa and Inovia Capital, and notable angel investors like Hermann Hauser (co-founder of Acorn and Amadeus Capital), Stan Boland (formerly of Icera, NVIDIA, Element 14, and Five AI), and Amar Shah (co-founder of Wayve). In total, Fractile has raised $17.5 million (£14 million) to date.

Founded in 2022 by Walter Goodwin, a 28-year-old AI PhD, Fractile has introduced a novel approach to AI chip design. This approach promises significant performance enhancements for advanced AI models in deployment.

The Current Landscape of AI Hardware

Today's AI companies are in a fierce race to develop, train, and deploy the most advanced foundational models. This requires substantial computational resources, but they all rely on similar hardware. These existing chips, while well-optimised for training large language models (LLMs), fall short during inference – the process of running live data through a model to generate results. The challenges include:

• High Costs: Running AI models at scale on conventional hardware is expensive, as the hardware often operates below its theoretical capacity due to inefficiencies
• Performance Constraints: The lag in inference caused by transferring model parameters from memory to processors limits real-time performance
• Future Limitations: The continuous advancement of conventional chips is hindered by heat generation, restricting the scalability of AI models
• Lack of Differentiation: Without novel hardware, AI model providers struggle to create significant competitive advantages in inference

Fractile has chosen a revolutionary path, opting to fundamentally transform computational operations and develop entirely new chips from the ground up. This approach promises breakthrough performance across a variety of AI models, both current and future.

Fractile aims to achieve remarkable performance improvements in AI model inference, targeting speeds 100 times faster and costs 10 times lower than current systems. By employing in-memory computing, Fractile eliminates the need to transfer model parameters between memory and processors, integrating computational operations directly into the memory. This method ensures compatibility with existing leading-edge silicon foundry processes.

Fractile's system also significantly reduces power consumption, aiming for 20 times the Tera Operations Per Second per Watt (TOPS/W) compared to other systems. This efficiency enables serving more users simultaneously and delivering faster responses, enhancing user experience.

Fractile's hardware allows AI model providers to leverage current models for superior performance, enabling advanced features like reasoning in language models. Faster speeds facilitate cost-effective recursive queries, chain-of-thought prompting, and tree searches, improving the quality of AI-generated responses. This leap in performance extends to various applications, from drug discovery to climate modelling and video generation.

Fractile has assembled a top-tier team, including senior hires from NVIDIA, ARM, and Imagination. The company has filed patents for its key circuits and unique in-memory compute approach. Fractile is in discussions with potential partners and anticipates forming partnerships before launching its first commercial AI accelerator hardware. The recent funding will support team expansion and accelerate the development of its initial product.

In summary, Fractile’s innovative AI chip design promises to revolutionise AI performance, offering significant improvements in speed, cost, and efficiency. With its groundbreaking approach and strong support from investors, Fractile is well-positioned to make a substantial impact in the AI hardware industry.

Dr Walter Goodwin, CEO and Founder of Fractile: “In today’s AI race, the limitations of existing hardware - nearly all of which is provided by a single company – represent the biggest barrier to better performance, reduced cost, and wider adoption. Fractile’s approach supercharges inference, delivering astonishing improvements in terms of speed and cost. This is more than just a speed-up – changing the performance point for inference allows us to explore completely new ways to use today’s leading AI models to solve the world’s most complex problems. We’re thrilled to have raised our funding from investors with a wealth of experience in the AI and chip industries, continue to grow our world-class team and further our technological development and partnerships.”

John Cassidy, Partner at Kindred Capital: “AI is evolving so rapidly that building hardware for it is akin to shooting at a moving target in the dark. It’s a major technical challenge but, with the AI inference chip market projected to be worth $91B by 2030, there’s also huge growth potential. Because Fractile’s team has a deep background in AI, the company has the depth of knowledge to understand how AI models are likely to evolve, and how to build hardware for the requirements of not just the next two years, but 5-10 years into the future. We’re excited to partner with Walter and the team on this journey.”

Andrea Traversone, Managing Partner at NATO Innovation Fund: “AI is one of the most transformational technologies of our time. Staying at the forefront of AI innovation is key to building a secure future and advancing defence, security and resilience across the Alliance. We are thrilled to be supporting Fractile, a company whose computing technology can enhance our collective AI capabilities by enabling computing power to run faster, more efficiently and sustainably.”

Sam Harman, Head of Deep Tech at Oxford Science Enterprises: “Fractile has developed a radically innovative approach to solve one of the AI sector’s biggest challenges. We’re delighted to be supporting them and look forward to seeing the company accelerate its development while contributing to the UK’s growing reputation as an amazing place to build an AI company.”

Stan Boland, angel investor: “There’s no question that, in Fractile, Walter is building one of the world’s future superstar companies. He’s a brilliant AI practitioner but he’s also listening intently to the market so he can be certain of building truly compelling products that other experts will want to use at scale. To achieve this, he’s already starting to build one of the world’s best teams of semiconductor, software and tools experts with track records of flawless execution. I’ve no doubt Fractile will become the most trusted partner to major AI model providers in short order.”