Fractile banked €208M ($220M) in a fresh funding round led by Accel. The capital injection also saw participation from Founders Fund. The London-based company is developing a new class of processors designed specifically to run large language models (LLMs) more efficiently than existing industry-standard hardware.
The startup focuses on overcoming the "memory wall"—the bottleneck that occurs when moving data between a processor and its memory—which currently limits the speed of AI inference. Fractile claims its architectural approach can execute AI workloads significantly faster and at a lower power cost than traditional GPUs by performing computations directly where the data resides.
This substantial investment follows a period of intense competition in the AI silicon space, where European hardware companies are seeking to provide alternatives to established US giants. The company intends to use the funds to scale its engineering efforts and bring its first commercial chips to market as demand for high-performance AI infrastructure continues to outpace supply.
As specialized AI workloads become the norm, the concentration of capital into European chip architecture suggests a growing regional ambition to secure the underlying compute layer of the generative AI stack.
Originally reported by Tech Funding News.