AI Designs Future Chips Faster, Cheaper

Artificial intelligence is poised to accelerate the creation of advanced silicon chips, addressing long-standing challenges in the semiconductor industry. Cognichip is developing a deep learning model designed to work collaboratively with engineers during the intricate and expensive process of chip design.

The company's CEO, Faraj Aalaei, stated that their technology can drastically cut chip development costs by over 75% and reduce project timelines by more than half. This advancement is crucial, as the lengthy design phase can take up to two years, with the entire process extending to five years, potentially leading to market obsolescence.

Cognichip has raised $93 million in total funding since its establishment in 2024, with a recent $60 million round led by Seligman Ventures. The company differentiates itself by training its AI model on specialized chip design data, including proprietary and licensed datasets, rather than general-purpose models.

While Cognichip has not yet revealed specific chips designed with its system or named its clients, its approach involves secure training procedures for chipmakers to use their own sensitive data. The company has also utilized open-source architectures like RISC-V in demonstrations, collaborating with students to design CPUs.

Cognichip faces competition from established players and well-funded startups in the AI chip design space. This period is characterized by a significant influx of capital into AI infrastructure, described as a "super cycle" for semiconductor and hardware companies.