AI-designed chips face patent infringement risks as autonomous tools emerge

AI's Role in Chip Design Expands

Artificial intelligence is taking on more complex tasks in chip design, moving beyond simple optimization. Companies like Synopsys, with its AgentEngineer tools, are developing AI capable of verifying, implementing, and planning analog chip designs. Synopsys plans general availability for its tools by year-end, with over 50 customer engagements already in progress.

Industry Adoption of AI Design Tools

Synopsys is not alone in this trend. Competitor Cognichip offers its ACI platform, which generates specifications, RTL, verification plans, and testbenches automatically. In China, Empyrean Technology claims an AI agent reduced a circuit layout task from four weeks to one. Even major AI labs are involved; OpenAI and Broadcom used AI to develop their custom inference chip, Jalapeño, in nine months.

Looming Patent Infringement Concerns

The rise of autonomous AI in chip design introduces a significant challenge: the potential for patent infringement. If an AI generates a design that incorporates existing patented intellectual property, and that design is used in mass-produced chips, it could lead to costly legal disputes. This issue is particularly complex given the industry's heavy reliance on licensing processor architectures, controllers, and memory technologies rather than designing every component from scratch.

Impact on the Chip Design Ecosystem

Experts like Domenec Forte and Simon Moore highlight this as a looming problem for the AI chip design market. The chip industry thrives on intellectual property, with companies like Arm generating substantial revenue from licensing and royalties. Synopsys also sells a portfolio of "silicon-proven" IP alongside its tools. The unclear legal implications of AI-generated designs could disrupt established IP licensing models and create uncertainty for manufacturers.