Normal Computing raises $50M to solve the AI energy crisis with thermodynamic chip technology

Normal Computing Secures $50 Million to Address AI Energy Consumption Crisis

Normal Computing has raised $50 million in funding led by Samsung Catalyst to tackle the soaring energy demands of AI chips, bringing its total capital raised to more than $85 million2

2

. The Series B round attracted new investors including Galvanize, Brevan Howard Macro Venture Fund, and ArcTern Ventures, alongside existing backers Celesta Capital, Drive Capital, Eric Schmidt's First Spark Ventures, and Micron Ventures1

1

. Founded in 2022 by former engineers and scientists from Google Brain, Google X, and Palantir, the startup is positioning itself at the intersection of two critical challenges facing the semiconductor industry: the escalating cost and power consumption of AI chips.

Electronic Design Automation Platform Gains Traction with Semiconductor Companies

CEO Faris Sbahi revealed that the company's software platform for semiconductor companies is already being used by more than half of the top 10 semiconductor firms by revenue1

1

. The Normal EDA platform addresses one of the industry's most pressing issues: the rising complexity of designing advanced AI chips, where even minor errors can trigger expensive delays and rework. Modern AI chips pack in tens of billions of transistors to support today's frontier models and can cost more than $500 million to develop before a single unit ships1

1

. The chip design software uses a frontier AI technique called auto-formalization, combining large language models with formal logic to help engineers design, optimize and prove the correctness of their silicon designs2

2

.

Thermodynamic Computing Chip Architecture Promises 1,000x Efficiency Gains

Beyond its commercial software business, Normal Computing is developing energy-efficient AI chips based on a radically different approach. The company has already achieved tape-out of its prototype thermodynamic computing chip, the CN101, which represents the world's first thermodynamic computing chip2

2

. Unlike traditional GPUs that consume massive amounts of energy fighting the inherent randomness of physical systems to maintain rigid "0" or "1" states, Normal's thermodynamic computing chip architecture harnesses thermal dynamics and treats nature's randomness as a feature rather than a bug2

2

. This approach aims to deliver 1,000-times energy efficiency gains, initially focused on generative AI inference workloads.

Data Centers Face Energy Wall by 2030

"The mission of the company is to go after this so-called AI energy crisis," Sbahi told Fortune. "Data centers are expected to hit an energy wall around 2030, and most of the strategy now is to find new ways to acquire more energy -- but our position is to solve the problem in terms of the hardware that we're using"1

1

. Normal Computing joins a growing cohort of startups exploring alternatives to conventional AI hardware, including Unconventional AI, led by former Intel AI chief Naveen Rao, which raised a $475 million seed round in January led by Andreessen Horowitz and Lightspeed Ventures. The company is researching its thermodynamic chip architecture in collaboration with the U.K.'s Advanced Research and Invention Agency, known as ARIA. Suraj Bramhavar, director of ARIA's Scaling Compute program, noted that Normal's team "has taken a fundamentally unconventional approach and delivered working silicon in CN101," calling it "an exceptionally rare outcome for work this ambitious"2

2

. The current fundraise will focus on scaling Normal's commercial Electronic Design Automation platform, with the long-term goal of integrating into mainstream semiconductor design manufacturing.