AI Designs Chips for AI... Cognichip Raises $60 Million in Funding

Semiconductor design is an extremely complex task requiring enormous cost and time. Nvidia's latest GPU, Blackwell, requires placing 104 billion transistors, and the design phase alone can take up to two years.

Cognichip aims to solve this problem by building deep learning models that assist engineers. Through this technology, the goal is to cut chip development costs by more than 75% and reduce development time by more than half.

Cognichip recently raised $60 million in new funding led by Seligman Ventures. Intel CEO Lip-Bu Tan participated in this round, bringing the company's total cumulative funding to $93 million.

Cognichip's core competitive advantage is using domain-specific models trained on chip design data instead of general-purpose LLMs. Given the security-sensitive nature of the semiconductor industry, the company employs the following strategies:

• Utilizing synthetic data and data licensed from partner companies
• Developing secure training procedures to prevent exposure of companies' proprietary data
• Validating models using open-source architectures such as RISC-V

The company currently competes not only with established players such as Synopsys and Cadence Design Systems, but also with emerging startups such as ChipAgents and Ricursive.