OpenAI Teams Up With Synopsys on GPT-Synopsys: AI Operating Chip Design Software

Teaming Up on a Specialized Model

On September 30, EDA giant Synopsys and OpenAI announced a multi-year partnership. Together they’ll develop a specialized model, GPT-Synopsys, trained specifically to operate Synopsys’s electronic design automation (EDA) tools.

EDA software is essential to chip design — engineers draw circuits, run simulations, and verify timing with these tools. Until now, AI in this space mostly stayed at the advisory level: suggestions, scripts. This time the goal is different: let AI agents actually get their hands on the software and use it as their own.

Per the joint statement, engineers only need to state design goals — power, performance, and area (PPA) optimization, say, or timing closure and verification signoff. The agent handles the rest: running tools, reading results, modifying designs, iterating — then handing results back to engineers for review.

How It Runs, How the Money Splits

The model runs on OpenAI’s infrastructure and will plug into Synopsys.ai and Synopsys’s agent platform Autopilot. Synopsys also says GPT-Synopsys will be compatible with customers’ own agent frameworks, so enterprises can wire it into their workflows.

On the business side, OpenAI is buying EDA tool training subscriptions from Synopsys to train the model on. The two also signed a revenue-sharing arrangement that, per Reuters, ties payouts to the actual design improvements the model delivers for customers. They’ll jointly take the product to global customers; no launch date or pricing announced.

Data security is the chip industry’s most sensitive nerve. Synopsys promises customer design data won’t be used to train models. Data is encrypted in transit and at rest, with configurable retention, auditing, and permissions. Outputs still go through traditional physical signoff tools — the mandatory hard check before tapeout, which no AI output can bypass.

The Market Reacted Immediately

The same day, Synopsys announced another multi-year deal: a $1 billion partnership with Amazon covering custom chips, AI products, and cloud infrastructure.

Stacked together, Synopsys raised FY2027 revenue guidance to $11.0–11.2 billion, above the $9.715 billion midpoint target for FY2026. Non-GAAP EPS guidance also rose to $19.04–19.12, versus $15.07 expected for FY2026.

The stock jumped 12.78% on Thursday to close at $490.54. Markets voting with real money: Wall Street will buy the story of AI agents moving into industrial software.

The Signal: Agents Moving Into Professional Software

Chip design is famously long-cycle and high-barrier — a complex chip takes months from design to verification. Reuters quotes OpenAI co-founder Greg Brockman: the partnership aims to cut design cycles by weeks or even months. Synopsys CEO Sassine Ghazi stressed that traditional signoff tools remain the physical guardrails; AI works inside them.

The signal: AI agents are moving from the “chat box” into professional software. EDA tools are complex, long-process, and near-zero-tolerance — among the hardest scenarios for agent deployment. If GPT-Synopsys works, it proves agents can operate professional tools in industry, and more sectors will follow.

Both sides say early technical engagement with top semiconductor customers is underway, without naming names. When the model launches and what it costs — still no word.