Synopsys and OpenAI announced a partnership on September 30, 2026, to develop GPT-Synopsys, a specialized artificial-intelligence model for semiconductor design. The companies say the system will combine OpenAI’s models with Synopsys’ electronic-design-automation tools and chip-engineering expertise.
The agreement is aimed at tasks that range from interpreting circuit descriptions to helping engineers place billions of transistors on a chip. It does not eliminate conventional verification: Synopsys says established physics-based tools will still be used to check whether designs meet manufacturing and performance requirements.
What the companies announced
In its September 30 release, Synopsys described GPT-Synopsys as a domain-specific model trained and tuned for chip-design work. The company said the partnership will integrate AI reasoning with engineering workflows rather than replace its existing sign-off and validation products.
Synopsys also identified OpenAI as a preferred partner for the project. The companies did not publish a general availability date, detailed performance results or a final customer price in the announcement.
How the commercial arrangement works
Reuters reported on September 30 that OpenAI will pay Synopsys a subscription fee for training and that the companies will share revenue from customer use according to the improvement the specialized model delivers. The exact financial terms were not disclosed.
That structure differs from a simple software-licensing deal. It links at least part of the partnership’s economics to adoption and measured value, while leaving important details about benchmarks and customer contracts undisclosed.
Why chip design is a difficult AI problem
Modern semiconductor development involves many connected stages: architecture, verification, logic synthesis, physical layout, power analysis and manufacturing sign-off. A change that improves one metric can create problems elsewhere, so suggestions must be checked against strict engineering constraints.
A specialized model could help engineers search a larger design space, generate or review code-like hardware descriptions and automate repetitive steps. But a plausible AI suggestion is not enough to manufacture a working chip. Traditional simulation, formal verification and physical-sign-off tools remain necessary to catch errors and prove that a design can be built.
What is known — and what is not
The announcement confirms the partnership and model name, but it does not show how GPT-Synopsys performs on independent evaluations. It also does not specify which Synopsys products will receive the technology first or when outside customers can test it.
Synopsys separately told investors that it expects about 15% revenue growth in its 2027 fiscal year, according to Reuters, but that company outlook should not be treated as a measure of GPT-Synopsys performance. The model is one part of a broader strategy that includes established design software and other AI-assisted engineering products.
For now, the project is best understood as a confirmed product-development partnership rather than a finished tool available for purchase. AskNovus will track product releases, customer access and independent evaluations in Technology, Artificial Intelligence and Business.
Featured image: Alexandre Debiève via Unsplash.
Sources
- Synopsys, September 30, 2026
- Synopsys investor release, September 30, 2026
- Reuters, September 30, 2026