As artificial intelligence moves beyond answering simple questions towards performing complex, multi step tasks, the demand for faster and more affordable computing is becoming one of the technology industry’s biggest challenges. British semiconductor startup Fractile is positioning itself at the centre of this shift, with an ambitious plan to redesign the hardware powering advanced AI systems. The company has raised $720 million in a major Series C funding round, reaching a valuation of approximately $7 billion.
The investment represents a significant milestone for the London and Bristol based company, which is developing specialised processors designed to accelerate AI inference. The round was led by Thrive Capital, Founders Fund, Lightspeed and Redpoint Ventures, with participation from Google’s CapitalG, BlackRock, DST Global, the UK’s Sovereign AI fund, the British Business Bank, Shopify Ventures, Balderton and several existing investors.
Building Chips for the Next AI Revolution
Founded in 2022 by Walter Goodwin, Fractile is focused on a part of artificial intelligence known as inference. This is the process through which an already trained AI model generates responses, writes code, analyses information or completes tasks.
While much of the AI hardware industry has concentrated on building powerful systems for training models, inference is becoming increasingly important as millions of people and businesses use AI applications every day.
As models become more capable, they also require greater computing resources to generate responses. Longer reasoning processes, AI agents and advanced coding applications can involve enormous numbers of calculations and data transfers.
Fractile believes existing chip architectures are struggling to keep pace with these requirements. Its strategy is to develop processors specifically designed for the growing demands of AI inference.
Rethinking the Relationship Between Memory and Computing
A central part of Fractile’s technology is its approach to connecting memory and computing components.
Conventional AI processors rely on separate memory and processing units. Data must repeatedly travel between these components, creating a bottleneck that can limit performance and increase energy consumption.
Fractile is developing an architecture that places memory much closer to computing components, including designs that stack DRAM directly on top of logic. By reducing the distance data needs to travel and increasing memory bandwidth, the company aims to improve the speed at which AI models generate outputs.
The approach is intended to support large language models, complex reasoning systems and applications that require substantial amounts of information to remain accessible during processing.
The company has reported progress in extracting substantially higher speeds from DRAM memory. Its ultimate objective is to deliver AI hardware that can operate at much higher performance levels without requiring proportionately higher infrastructure costs.
Challenging Established Chipmakers
Fractile is entering a market dominated by established semiconductor companies, particularly Nvidia, whose processors have become essential to modern AI infrastructure.
However, the growing demand for inference is creating opportunities for specialised chip developers. Different AI workloads require different combinations of processing power, memory capacity, bandwidth and energy efficiency.
Fractile is taking a full stack approach, working across chip architecture, software, systems engineering and foundational AI research. This allows the company to optimise more parts of the technology rather than depending entirely on conventional processor designs.
Funding to Accelerate Commercial Launch
The latest investment will help Fractile expand its engineering capabilities and accelerate the development of its first commercial chips and complete computing systems. The company has indicated that it aims to ship its first chips and systems in 2027.
The funding also strengthens Britain’s position in the global AI hardware race. The participation of the Sovereign AI fund and British Business Bank reflects growing interest in developing domestic capabilities in strategically important technologies.
For Fractile, the challenge now is to turn its ambitious architecture into commercially deployable hardware. If successful, its technology could help make advanced AI applications faster, more affordable and accessible to a wider range of businesses.
With substantial investor backing and a growing market for inference computing, Fractile is seeking to establish itself as a major contender in the next generation of AI infrastructure.
