Normal Computing develops AI-accelerated software and physics-based hardware to improve the efficiency of AI infrastructure and semiconductor design. The company builds two main offerings: Normal EDA, an AI-accelerated co-design platform for silicon engineering teams, and Normal ASICs, physics-based application-specific integrated circuits using thermodynamic computing that claim to deliver up to 1,000x more intelligence per unit of energy.
Founded in the USA by former members of Google Brain and Google X, the company operates offices in New York City, San Francisco, London, and Copenhagen. Its team draws from leading technology and research organizations including Meta, Palantir, IBM Research, UC Berkeley, NVIDIA, and Intel. Normal Computing works in partnership with major semiconductor design and manufacturing institutions.
The company's technical focus spans AI for the physical world, probabilistic and quantum AI, machine learning frameworks, hardware-software co-design, and silicon engineering. Its core premise is that general-purpose computing architectures underutilize the physical potential of hardware, and that aligning design with the intrinsic properties of physical systems can transform efficiency across the semiconductor industry.
