Normal Computing Corp. raises $50M in Series B

Normal Computing Corp., an emerging player in AI chip design, has secured $50 million in a Series B funding round led by Samsung Catalyst. The round also attracted investments from Galvanize, Brevan Howard Macro Venture Fund, and ArcTern Ventures, while existing investors such as Celesta Capital, Drive Capital, Micron Ventures, and Eric Schmidt’s First Spark Ventures continued their support. The latest capital infusion brings the company's total funding to over $85 million.

Normal Computing is addressing a pivotal challenge in the AI industry: the rising energy consumption of artificial intelligence applications. As AI models become more complex, they impose substantial power demands on conventional graphics processing units (GPUs). CEO Faris Sbahi warns of an imminent "energy wall," where large-scale AI computations may become untenable due to prohibitive energy requirements. To counter this, the company is redesigning silicon chip architecture and developing a new processor type that aligns with physical laws, rather than resisting them.

A cornerstone of Normal's strategy is its proprietary electronic design automation (EDA) platform, currently utilized by five of the world's top ten semiconductor design firms. This platform leverages advanced AI techniques known as "auto-formalization" to streamline chip design. By employing large language models integrated with formal logic, the platform aids engineers in optimizing and verifying silicon designs. The approach promises to cut development cycles from several years to months, while also enhancing efficiency, potentially realizing 1,000-times gains, according to Sbahi.

In addition to its redesign efforts, Normal is pioneering thermodynamic computing, an approach that embraces the natural randomness of physics to perform computations. Traditional silicon chips expend significant energy to maintain binary states. Normal's technology, however, allows such systems to utilize thermal dynamics, making calculations more energy efficient. The company's debut product, the CN101 chip, represents the first step in achieving this transformative energy efficiency.

Normal's collaboration with the UK's Advanced Research and Invention Agency (ARIA) highlights the strategic importance of its work. Suraj Bramhavar, director of ARIA's Scaling Compute program, emphasizes the potential for Normal's technology to facilitate a substantial leap in chip performance, moving beyond incremental improvements. The firm’s next milestones likely involve advancing its thermodynamic computing architecture and fostering partnerships to scale its innovative solutions across the AI chip sector.