Cognichip Brings AI-Driven Design to Semiconductors

Cognichip, builders of ACI® Artificial Chip Intelligence, led by Faraj Aalaei, Ehsan Kamalinejad, Simon Sabato, and the team, has raised $60 million in new funding led by Seligman Ventures, with notable participation from Intel CEO Lip-Bu Tan, who will be joining Cognichip’s board. Umesh Padval, a managing partner at Seligman, will also join the board.

Cognichip is developing a deep learning model to assist engineers in designing next-generation chips, tackling a long-standing industry challenge: chip design is complex, costly, and slow. Advanced chips can take 3–5 years to reach production, with up to two years spent on design alone. For context, Nvidia’s Blackwell GPUs contain over 100 billion transistors, highlighting the scale of the challenge.

Cognichip CEO Faraj Aalaei notes that chip development cycles are so long that market shifts can render designs obsolete before launch. His goal is to bring AI-driven productivity, similar to what software engineers use, into semiconductor design.

“These systems are now intelligent enough that, with guidance, they can produce high-quality outputs,” Aalaei said. The company claims its technology can cut chip development costs by over 75% and reduce timelines by more than half.

However, Cognichip has yet to showcase a chip fully designed using its system and has not disclosed the customers it claims to be working with since September.

Cognichip’s edge lies in using a custom model trained specifically on chip design data, rather than a general-purpose LLM. Accessing this data is challenging, as semiconductor IP is tightly guarded and lacks the open-source ecosystem seen in software.

To overcome this, the company has built its own datasets, including synthetic data, and licensed additional data from partners. It has also developed secure methods that allow chipmakers to train models on proprietary data without exposing sensitive information.

Where proprietary data is limited, Cognichip leverages open-source alternatives. In a recent demo, electrical engineering students used the model to design CPUs based on the open-source RISC-V architecture, demonstrating early capabilities.

The company competes with incumbents like Synopsys and Cadence Design Systems, as well as startups such as ChipAgents and Ricursive.

Amid a surge in AI infrastructure investment, investors see a broader semiconductor “super cycle”, one that could strongly benefit companies like Cognichip.