OpenAI and Synopsys build a model to operate chip-design tools
Greg Brockman says the multi-year partnership will help engineers explore designs faster; the companies have disclosed no launch date, benchmarks or named customers.
By RuntimeWire Staff · Published
Primary source: Aligned News - AI Intelligence
Why it matters
OpenAI is taking its models into the engineering systems behind AI hardware. The partnership's value will depend on verifiable improvements to chip-design workflows, not the model label.
On September 30th, Greg Brockman (@gdb), OpenAI's president and co-founder, and Synopsys CEO Sassine Ghazi announced a multi-year partnership to develop GPT-Synopsys, a model designed to operate Synopsys' semiconductor design tools. The Synopsys announcement describes a product under development, with early customer engagements underway. It gives no launch date, benchmark results or customer names.
The announcement also appeared in:
Brockman framed the deal as a way for OpenAI to help improve the hardware that powers AI. "By helping them build better chips, we can build better AI and bring it to more people," he said in the announcement. That pitch puts OpenAI further into the tools and infrastructure behind AI, beyond the software people use to ask models for help. Brockman, formerly Stripe's CTO, helped establish OpenAI in 2015. The partnership applies OpenAI's models to a demanding engineering process with measurable outputs.
From prompts to engineering workflows
The planned system would let engineers specify goals such as improving power, performance and area, closing timing gaps or verifying a design. AI agents would run Synopsys tools, interpret their results, make changes and repeat the process until they reach outcomes for engineers to review. Synopsys says OpenAI will license its EDA software for model development. The planned service will run on OpenAI-hosted infrastructure and integrate with Synopsys.ai and the Synopsys Autopilot platform.
That workflow gives the partnership a more concrete job than a general-purpose chatbot answering chip-design questions. Electronic design automation tools help engineers create, test and verify chip designs; GPT-Synopsys is meant to use those tools directly. The hard part is keeping each automated step inside a process engineers can validate before a design advances toward manufacturing. The companies describe engineers reviewing the results, but have not published a benchmark showing how reliably the model completes a workflow or how much time it saves.
The timing also fits Synopsys' own move toward agents that execute longer design tasks. Synopsys has been developing AgentEngineer, its portfolio of AI systems for multi-step engineering workflows. GPT-Synopsys pairs that accumulated tool and domain expertise with OpenAI's models. Under the agreement, OpenAI and Synopsys will jointly develop and deliver GPT-Synopsys. OpenAI brings frontier AI and hosts the infrastructure, while Synopsys contributes its EDA tools and chip-design expertise. The companies will also collaborate on go-to-market work and share revenue, though neither has disclosed the commercial split or pricing.
The proof will be in the design results
For Brockman, the partnership adds a new kind of agent work to OpenAI's expanding product push. Just a day earlier, OpenAI introduced Dots, always-on agents that work across connected apps, as RuntimeWire reported. GPT-Synopsys takes the same broad idea - models carrying out multi-step tasks - into a specialized environment where success depends on what the engineering tools report, not just whether an answer sounds plausible.
Ghazi brings a different perspective to the collaboration. He has spent more than 25 years at Synopsys, held leadership roles across engineering, product development, sales and operations, and became CEO in 2024. His challenge is to expand how customers use Synopsys' design tools while preserving the trust engineers need in results that may ultimately inform manufactured silicon. OpenAI and Synopsys say the joint service will bundle compute, the model and tool licenses, and that customer design data will not be used to train the model. Those protections are promises in the announcement; the companies have not yet published product terms for customers to evaluate.
OpenAI and Synopsys say early technology engagements are underway with semiconductor customers. They do not name those customers, say what designs or workflows are being tested, or report a completed chip-design outcome. The announcement's claims about capability, timing and availability are forward-looking. Until the partners show measured results against an engineering baseline, "faster chip design" remains the goal of the collaboration, not a demonstrated product result.
The commercial structure gives both companies a stake in getting from research to a service: OpenAI licenses Synopsys tools, Synopsys integrates the model into its own software, and the partners plan to share revenue. That arrangement also leaves practical questions for customers, including who gets access first, how the service will be priced and how engineers can inspect or reproduce an agent's changes. The answers will determine whether GPT-Synopsys becomes an everyday part of chip development or stays a compelling demonstration of what a model might do.