OpenAI and Synopsys announced on September 30, 2026 a multi-year agreement to develop GPT-Synopsys, an AI model specialized in chip design. The system is meant not only to suggest next steps, but to operate Synopsys electronic design automation (EDA) tools the way an experienced engineer would.
What the deal covers
According to the official Synopsys release, OpenAI will license Synopsys EDA tools to train and tune the model. The companies will collaborate on R&D, joint go-to-market work, and a shared-revenue framework when customers use the product.
The model is expected to run on OpenAI-hosted infrastructure, integrate with Synopsys.ai and Synopsys Autopilot, and interoperate with customer agent harnesses. Engineers would delegate goals — power, performance and area (PPA) optimization, timing and verification closure — while agents run tools, interpret results, apply changes and iterate toward verified outcomes for human review.
Why it matters
Modern chip design is a maze of trade-offs among performance, power and silicon area. General-purpose models already connect to EDA tools. The announced leap is to make a frontier model a native expert in that loop.
Synopsys president and CEO Sassine Ghazi said semiconductor engineering needs faster design without sacrificing PPA or first-time-right silicon. OpenAI president and co-founder Greg Brockman said the company is using its most advanced technology to improve the systems that power AI itself: “By helping them build better chips, we can build better AI and bring it to more people.”
Reuters reported that OpenAI will pay Synopsys a training subscription to learn the tools, and that future revenue will depend on how much the model improves a given chip design.
What is not ready yet
There is no public date for broad commercial availability. The companies say early engagements with semiconductor customers are underway. Customer design data will not be used to train the model, according to the release, and is to be encrypted in transit and at rest, with retention, audit and permission controls.
The announcement does not replace engineers: the described workflow is delegation with human review, not a chip leaving the fab on its own.
Sources
Transparency: This content was created, edited or reviewed with the assistance of artificial intelligence. Information was cross-checked with public posts on X and sources available on the internet. Consult the original sources for the full context.
Por GeekikiBot