OpenAI and Synopsys Are Building an AI That Designs Chips

by

OpenAI is teaming up with Synopsys, the company whose software sits behind a huge share of the world’s chip design, to build an AI model that does chip-design work itself. The model, called GPT-Synopsys, will be trained and tuned specifically for the semiconductor business, the two companies announced Wednesday at a Synopsys investor summit.

Chip design today starts with engineers describing circuits in a code-like language and ends with the painstaking puzzle of arranging billions of transistors on a sliver of silicon. GPT-Synopsys is meant to work across that whole arc. Instead of operating on its own, the model will learn to use Synopsys’s own design tools — helping engineers weigh trade-offs and squeeze better performance, power efficiency, and density out of their layouts.

The promise is speed. OpenAI co-founder Greg Brockman said in an announcement video that the goal is “to be able to shave off weeks, months from the design process and to bring more chips to the world.” With the semiconductor industry stuck in a loop of shortages and surging AI-driven demand, faster design cycles would be a genuine unlock — for giant chipmakers and for small teams that can’t afford drawn-out engineering timelines.

The money side is built around results. OpenAI will pay Synopsys a training subscription fee for the privilege of learning how to use its tools. Once customers start using the finished product, the two companies will split revenue based on how much the model actually improves the design of a chip — a performance-linked cut rather than a flat license fee. Synopsys CEO Sassine Ghazi told Reuters the deal is deliberately structured so it won’t cannibalize Synopsys’s existing business, calling it upside driven by extra customer value.

One important caveat: this isn’t AI designing chips unsupervised. Ghazi said the model’s output will still be checked by Synopsys’s traditional verification tools — the same “sign-off” process engineers rely on today. “The model needs these guardrails in order to check the physics,” he said. “The need for validating with the highest level of fidelity… is essential.”

Investors liked what they heard. Ghazi also raised Synopsys’s fiscal 2027 revenue growth outlook to 15%, well above the roughly 11% analysts expected, and shares rose as much as 7% after the announcement.

Sources:

Leave a Comment