Synopsys, Inc. (SNPS.US) partners with OpenAI to develop AI models for chip design; revenue-sharing model drives stock up as much as 7%.

Chip design software maker Synopsys (SNPS.US) announced on Wednesday that it has entered into a partnership agreement with OpenAI. The two parties will jointly develop AI models for the chip business and share related revenue. The model, named GPT-Synopsys, will be specially trained and tuned to take on chip design tasks. The chip design process typically begins with describing chip circuits in a code-like language, and then extends to how to lay out billions of transistors on a tiny silicon die. OpenAI co-founder Greg Brockman said in a video announcing the news that OpenAI's model will learn how to use Synopsys, Inc.'s software tools to help engineers handle various trade-offs and optimizations in the chip design process, "thereby shortening the design process by weeks, months, and bringing more chips into the world." Synopsys, Inc. announced the deal at an investor summit. Company CEO Sassine Ghazi said the same day that Synopsys, Inc. currently expects 15% revenue growth in fiscal 2027, higher than the 11.19% analyst estimate shown by LSEG data. After the partnership and forecast news were announced, Synopsys shares rose as much as 7%. In an interview with Reuters, Ghazi said OpenAI will pay Synopsys, Inc. a training subscription fee to learn how to use Synopsys, Inc.'s tools. When Synopsys, Inc. customers use the product, Synopsys, Inc. and OpenAI will share revenue based on the degree to which the model improves chip design. Ghazi said: "We designed the agreement structure so that it will not cannibalize our business. Given that we deliver more value to customers, this will become an upside factor for our business." However, Ghazi also emphasized that the work of GPT-Synopsys will still be reviewed by Synopsys, Inc.'s tools using traditional computing technology to verify whether the chip can work properly. He said: "The model needs these guardrails to check physics. Verification at the highest fidelity, which is what we call signoff or ground truth, is crucial." The core significance of this deal is that AI is further penetrating from general-purpose large models into the EDA (electronic design automation) sector upstream in the semiconductor industry chain. Chip design is highly complex, involving multiple stages from circuit description, logic synthesis, placement and routing, to physical verification, and engineers need to repeatedly weigh performance, power, area, and cost. If an AI model can learn and invoke Synopsys tools to automatically complete part of the optimization and trade-offs, it could in theory significantly shorten the design cycle and reduce trial-and-error costs. From a business model perspective, the partnership between Synopsys, Inc. and OpenAI is not simple software licensing, but a combination of "training subscription fee + revenue sharing." OpenAI first pays a fee to learn how to use Synopsys tools; when Synopsys, Inc. customers use GPT-Synopsys, the two parties then share revenue based on the degree to which the model improves chip design. This pay-for-performance model helps reduce customers' concerns about the uncertain value of AI tools, and also binds OpenAI's interests to the actual benefits of Synopsys, Inc. customers. However, the two parties did not disclose key details such as the revenue-sharing ratio, exclusivity arrangements, and how customer design data will be used, which will be key areas of market attention going forward.