British startup Fractile raises $720M to build specialized AI chips

Securing a stake in AI hardware

Fractile, a British startup, has successfully closed a $720 million Series C funding round. This significant capital injection positions the company as a major player in the competitive market for purpose-built AI chips. The round was led by a group of prominent investors including Thrive Capital, Founders Fund, Lightspeed Venture Partners, and Redpoint.

A key participant in this funding is Britain’s Sovereign AI Fund. The British government stated that this investment is intended to ensure the U.K. has a greater stake in the hardware that will underpin its future economy. By backing domestic hardware development, the government hopes to reduce reliance on existing supply chains for critical computing infrastructure.

This funding places Fractile among the top-funded startup players in the AI chip space. It joins other notable companies like Axelera, Etched, and OLIX in the race to create specialized silicon. The goal is to provide a viable alternative to general-purpose hardware for the growing demands of artificial intelligence.

Solving the memory bottleneck

One of the primary challenges in modern AI is the tradeoff between memory and capacity. As companies move toward agentic AI—systems that can perform long chains of tasks—they need to run large models without sacrificing speed. Current hardware often struggles to keep up with these demands because moving data between memory and the processor creates significant delays.

Fractile's hardware design addresses this by focusing on DRAM, which is the most common type of random-access memory. The company plans to stack DRAM directly on top of each chip. This physical architecture is intended to bypass the usual limitations that occur when trying to balance how much data a chip can hold with how fast it can process that data.

By optimizing this physical layout, Fractile aims to make it easier for developers to run large models in production. This is particularly important for reducing the cost of inference, which is the process of using a trained AI model to generate outputs. Efficient inference is essential for making AI tools affordable and usable at scale for businesses.

A full-stack software approach

Fractile does not intend to sell hardware in isolation. The company is pursuing a full-stack approach that combines custom silicon with specialized software. This includes a machine learning compiler, which is a tool that translates high-level AI models into instructions that the hardware can execute efficiently.

By providing both the chip and the compiler, Fractile aims to maximize the speed at which large language models (LLMs) are executed. This integrated strategy allows the company to frame itself as a frontier AI company rather than just a hardware manufacturer. It ensures that the software is perfectly tuned to the unique capabilities of their specific chip design.

This holistic approach is designed to provide a smoother experience for developers. Instead of having to manually optimize code for new hardware, users can rely on the compiler to handle the complexities. This could significantly lower the barrier to entry for companies wanting to deploy high-performance AI systems.

Valuation and production timeline

The financial scale of this round reflects the high expectations for Fractile's technology. While the specific valuation was not officially confirmed in the funding announcement, reports from the Financial Times and Reuters suggest it is over £5 billion. This figure could potentially translate to $7 billion, highlighting the massive interest in AI infrastructure.

Despite the large amount of capital raised, the company acknowledges that the development process is still in its early stages. CEO Walter Goodwin noted in a LinkedIn post that "there is still a lot of work ahead" for the team. The focus remains on moving from design and research into physical production and deployment.

The company has set a clear target for its next major milestone. The new funding will provide the resources needed for the startup to deliver its initial chips and systems during the coming year. If successful, these shipments will mark the transition from a well-funded startup to a functional provider of AI hardware for the global market.