Overview
- The companies announced the multi‑year partnership and the GPT‑Synopsys model on Sept. 30, 2026, saying the model will be specialized to run electronic design automation (EDA) tools for semiconductor design.
- GPT‑Synopsys is being built to operate Synopsys’ EDA tools directly, run multi‑step agent workflows that interpret tool outputs and iterate designs, and will run on OpenAI‑hosted infrastructure integrated with Synopsys.ai and the Autopilot platform.
- Commercial terms include a subscription training fee paid by OpenAI to Synopsys and a revenue‑share arrangement that ties payments to how much the model improves customer chip designs, and Synopsys raised its fiscal 2027 revenue outlook after the announcement.
- The companies say customer design data will not be used to train the model and will be protected with encryption, retention, audit and permission controls, and they emphasize that traditional sign‑off verification will remain the ground truth for manufacturable designs.
- Early technology engagements and unnamed pilots are already under way, the deal builds on Synopsys’ prior AI work and recent industry moves such as Nvidia’s investment and OpenAI’s chip efforts, and the partnership could shorten design cycles while carrying execution and disclosure risks because timing and independent performance metrics remain undisclosed.