Synopsys reaches agreement with OpenAI to jointly develop artificial intelligence models for chip design

Stephen Nellis

Reuters San Francisco, September 30 - Chip design software maker Synopsys (SNPS.O) said on Wednesday that it has reached an agreement with OpenAI to share revenue from artificial intelligence (link) models developed for the chip business.

The model will be called “GPT-Synopsys” and will be specially trained and fine-tuned to undertake chip design tasks—ranging from describing chip circuits in code-like languages to ultimately determining how to lay out billions of transistors onto a tiny silicon wafer.

OpenAI co-founder Greg Brockman said in a video announcing the news that OpenAI’s models will learn how to use Synopsys’s software tools to help engineers handle trade-offs and optimization problems in the design process, “thereby cutting design cycles by weeks or even months and introducing more chips to the world.”

Synopsys announced the collaboration at an investor summit. According to London Stock Exchange Group (LSEG) data, Synopsys CEO Sassine Ghazi also said at the summit that the company now expects fiscal 2027 revenue to grow by 15%, higher than analysts’ expectations of 11.19%. Following the news of the partnership and earnings forecast, the company’s share price rose as much as 7%.

In an interview with Reuters, Ghazi said OpenAI will pay Synopsys subscription fees for training to learn to use Synopsys’s tools. When Synopsys’s clients use the product, Synopsys and OpenAI will share revenue based on the model’s optimizations for chip design.

“We designed this agreement structure to ensure it does not cannibalize our business,” Ghazi said. “Since we are creating more value for customers, this will have a positive impact on our business.”

Ghazi said that the results generated by the GPT-Synopsys model must still be reviewed by Synopsys’s tools, which use traditional computing technology, to verify that the chips are functioning properly.

“The model needs these safety guardrails to verify physical properties,” Ghazi said. “Verifying with the highest fidelity—which is what we call ‘sign-off’ or ‘golden reference’—is crucial.”

(To facilitate non-English speakers, Reuters automatically translates its reports into several other languages. Since automated translations may contain errors or lack required context, Reuters does not guarantee the accuracy of automated translation texts, which are provided only for the convenience of readers. Reuters assumes no responsibility for any damage or loss caused by the use of the automated translation feature.)