Cognichip wants AI to design the chips that power AI, and just raised $60M to try

The most advanced silicon chips have accelerated the development of artificial intelligence. Now, Cognichip is betting that AI can return the favor.

Cognichip is building a deep learning model to work alongside engineers as they design new computer chips. The problem it aims to solve is one the semiconductor industry has faced for decades: chip design is enormously complex, extremely expensive, and slow.

The challenge of modern chip design

Advanced chips can take three to five years to go from conception to mass production. The design phase alone may last up to two years before physical layout even begins. For example, Nvidia’s latest Blackwell GPUs contain 104 billion transistors — an immense level of complexity that must be carefully engineered.

According to Cognichip CEO and founder Faraj Aalaei, by the time a new chip is ready, market conditions may have shifted enough to undermine the original investment thesis.

Aalaei’s goal is to bring the kind of AI tools that have accelerated software development into the semiconductor design process.

“These systems have now become intelligent enough that by just guiding them and telling them what the result is that you want, it can actually produce beautiful code,” Aalaei told TechCrunch.

He says Cognichip’s technology can reduce the cost of chip development by more than 75% and cut the timeline by more than half.

$60 million in new funding

The company emerged from stealth last year and on Wednesday announced it had raised $60 million in new funding led by Seligman Ventures. Intel CEO Lip-Bu Tan participated in the round and will join Cognichip’s board. Umesh Padval, managing partner at Seligman Ventures, will also join the board.

Since its founding in 2024, Cognichip has raised a total of $93 million.

However, the company has not yet unveiled a chip designed using its system and has not disclosed the customers it says it has been collaborating with since September.

Training AI on chip design data

Cognichip says its advantage lies in using a proprietary model trained specifically on chip design data, rather than relying on a general-purpose large language model.

Accessing domain-specific training data is particularly challenging in semiconductors. Unlike software developers, who often share large amounts of code publicly, chip designers closely guard their intellectual property, limiting the availability of open datasets that typically power AI coding assistants.

To address this, Cognichip has developed its own datasets, including synthetic data, and licensed additional data from partners. The company has also created procedures that allow chipmakers to securely train Cognichip’s models on their proprietary data without exposing it.

Where proprietary data is not available, the company has used open source alternatives. In a demo last year, Cognichip invited electrical engineering students at San Jose State University to participate in a hackathon. Teams used the model to design CPUs based on the open source RISC-V chip architecture, which is freely available for anyone to build upon.

Competitive landscape

Cognichip is entering a competitive market that includes established players such as Synopsys and Cadence Design Systems, as well as well-funded startups. ChipAgents closed a $74 million extended Series A in February, and Ricursive raised a $300 million Series A round in January.

Padval said the current surge of investment in AI infrastructure is the largest he has seen in 40 years of investing.

“If it’s a super cycle for semiconductors and hardware, it’s a super cycle for companies like [Cognichip],” he said.