OpenAI and Synopsys announce GPT‑Synopsys, a model that runs chip‑design tools
The two companies say GPT-Synopsys will operate Synopsys' chip-design software on its own, with engineers delegating goals and revenue shared with OpenAI.
OpenAI and Synopsys announced a multi-year agreement to build GPT-Synopsys, a model specialised to use Synopsys' electronic design automation tools for semiconductor design, according to their joint press release. Commentator Andrew Curran posted the news on X ahead of the announcement, saying OpenAI had trained a specialised chip-design model.
https://x.com/AndrewCurran_/status/2105368300889034905
The companies say engineers will be able to delegate design objectives. The system would run the tools, interpret results, make changes and iterate toward verified outcomes, according to the release. That describes the intended behaviour, and the companies have not published benchmarks or a public demo of it working on a real design.
The terms, as summarised by outlets that read the release, have OpenAI licensing Synopsys' tools to train the model and paying a training subscription fee. The companies will share revenue when Synopsys customers use the product, with the split tied to how much it improves chip design. No price or revenue figure was disclosed.
GPT-Synopsys will run on OpenAI-hosted infrastructure and connect to Synopsys.ai and its Autopilot platform, the companies say. They also say customer design data will not be used for model training. Early engagements with semiconductor customers are under way, according to the release, but no customer has been named.
Chip design has long leaned on automation, and OpenAI's hardware chief has said AI helped design its own Jalapeño chip. What is new here is a frontier lab and the leading design-software vendor jointly shipping an agent for the tools. Whether it beats existing approaches is not yet established.