Pat Gelsinger, the former CEO of Intel, has made a significant investment in the British AI chip startup, Fractile. Gelsinger shared the news on LinkedIn on January 22, 2025, although the financial details surrounding the investment remain undisclosed.
About Fractile: Innovating AI Inference with In-Memory Compute
Fractile, founded in 2022 by Dr. Walter Goodwin, a former PhD student at the University of Oxford’s Robotics Institute, aims to address a major bottleneck in artificial intelligence (AI) model inference. Goodwin observed that as AI development shifted from model training to model inference, existing hardware could not support the required scalability. This gap prompted him to create Fractile, which develops chips based on in-memory compute technology.
The company’s in-memory compute approach allows processors to run calculations directly in the computer memory, thus reducing power consumption while improving overall performance. This innovation is poised to make AI inference faster, more cost-effective, and scalable—essential for modern AI applications, particularly in large-scale environments.
Fractile’s Funding Journey and Strategic Partnerships
In July 2024, Fractile exited stealth mode after securing $15 million in seed funding. The funding round was co-led by Kindred Capital, Nato Innovation Fund, and Oxford Science Enterprises, with participation from various angel investors. Notably, entrepreneur Stan Boland, known for his leadership roles at Arm and Acorn Computers, also invested in the company and serves as a director.
Fractile’s success in raising substantial funding demonstrates growing interest and confidence in its innovative approach to AI hardware.
Pat Gelsinger’s Role and Vision for AI
Gelsinger’s investment in Fractile aligns with his belief in the transformative potential of AI. In his LinkedIn post, he highlighted the critical role hardware plays in scaling AI inference, noting that current AI models face significant challenges with cost, latency, and power consumption. Gelsinger emphasized that achieving advancements in AI will require much faster, cheaper, and more energy-efficient inference solutions.
Reflecting on his own graduate work at Stanford, Gelsinger noted that Fractile’s approach mirrors some of the ideas he explored during his studies. As part of his involvement, Gelsinger expressed excitement about advising Fractile’s team as they work to overcome the hurdles of AI hardware development.
Why This Investment Matters for the AI and Semiconductor Industries
Gelsinger’s investment is seen as a significant endorsement for Fractile’s innovative approach to AI chip development. By focusing on in-memory compute, Fractile has the potential to reduce the energy footprint of AI inference tasks, which has become a pressing issue in the tech industry. This move also highlights the broader shift towards specialized AI hardware designed to handle increasingly complex workloads.
With Gelsinger’s expertise and guidance, Fractile is positioned to make waves in the rapidly growing field of AI semiconductor technology.
Frequently Asked Questions (FAQ)
1. What is Fractile’s core innovation?
Fractile is developing AI chips based on in-memory compute technology, allowing processors to perform calculations directly in computer memory. This method reduces power consumption, enhances performance, and supports scalable AI inference.
2. Who is Pat Gelsinger, and why is his investment significant?
Pat Gelsinger is the former CEO of Intel, renowned for his leadership in the semiconductor industry. His investment in Fractile underscores his belief in the importance of specialized AI hardware to overcome current challenges in AI scaling, particularly around cost and power consumption.
3. What is in-memory compute, and how does it benefit AI inference?
In-memory compute involves performing calculations directly within the memory of the processor, rather than moving data in and out of memory during computation. This results in reduced energy usage, faster processing, and improved performance for AI inference.
4. How much seed funding has Fractile raised?
Fractile raised $15 million in seed funding in a round co-led by Kindred Capital, Nato Innovation Fund, and Oxford Science Enterprises, with additional support from angel investors.
5. What challenges does AI inference face, and how does Fractile address them?
AI inference is bottlenecked by existing hardware, facing issues such as high costs, latency, and energy inefficiency. Fractile’s in-memory compute chips aim to tackle these challenges, offering faster, more efficient, and cost-effective AI processing at scale.
