Synopsys and OpenAI reach AI chip design partnership, stock price surges over 7%

Chip design software manufacturer Synopsys has entered into a multi-year strategic partnership with OpenAI aimed at revolutionizing semiconductor design workflows, with both parties jointly developing AI models dedicated to chip design.

On September 30, Synopsys announced its collaboration with OpenAI at an investor summit, stating that both companies will co-develop AI models for chip business applications.

The model will be named GPT-Synopsys and will be specifically trained and fine-tuned to undertake chip design tasks. These tasks range from describing chip circuits in code-like languages to determining how to lay out billions of transistors on small pieces of silicon.

According to the agreement, OpenAI will pay Synopsys a subscription fee for tool licensing, and both parties will share revenue from chip design improvements based on customer usage of the product. Following the announcement, Synopsys shares rose by over 7% intraday before closing up 4.78%.

OpenAI co-founder Greg Brockman said that GPT-Synopsys will help engineers cut weeks or even months from the chip design cycle and accelerate more chips coming to market.

Synopsys CEO Sassine Ghazi emphasized that the structure of this cooperation agreement would not cannibalize existing business but would bring incremental value to the company.

According to details revealed by Ghazi in an interview with Reuters, OpenAI will initially pay training subscription fees to learn how to use Synopsys' Electronic Design Automation (EDA) tools.

Once GPT-Synopsys is officially launched to customers, both sides will share revenue proportionally according to the actual improvement that the model brings to chip design.

Ghazi stated that the agreement was structurally designed to ensure the new product does not compete with Synopsys' existing business. "This will be incremental to our business because we are providing more value to customers."

GPT-Synopsys will operate on OpenAI’s own infrastructure and be deeply integrated with Synopsys's Synopsys.ai and Autopilot agent platforms.

Both parties noted that leading semiconductor customers are already involved in early-stage technical collaboration. Customer design data will be protected with enterprise-grade security, not used for model training, and encrypted throughout transmission and storage.

The core philosophy behind GPT-Synopsys is to transform state-of-the-art AI models from general-purpose tools into domain experts for chip design.

As described by both parties, current agent-based AI technology can already connect a general model to EDA tools to run design workflows. The breakthrough in this partnership is enabling the model to truly operate EDA tools like a seasoned engineer—including interpreting tool outputs and optimizing designs iteratively.

Practically, engineers will be able to delegate design objectives such as power, performance, and area (PPA) optimization, timing closure, and verification to the agent. The model will autonomously operate tools, interpret results, implement modifications, and iterate until a verification result is ready for engineer review.

However, Ghazi emphasized that outputs from GPT-Synopsys still require verification by Synopsys' traditional computation validation tools. He said:

The model needs these guardrails to verify physical laws and be validated with the highest fidelity—what we call sign-off or ground truth, which is an indispensable process.

At the investor summit, Synopsys also raised its performance guidance. The company now expects its fiscal 2027 revenue to grow by 15%, while LSEG data shows previous analyst consensus at only 11.19%.

Ghazi said at the conference that the partnership with OpenAI will extend Synopsys's advanced design capabilities and broaden its market reach for increasingly complex chip designs.

On the OpenAI side, Greg Brockman positioned the partnership as improving AI’s own compute infrastructure with cutting-edge AI technology:

By helping engineers build better chips, we can build better AI and deliver it to more people.