TL;DR: OpenAI and Synopsys will co-develop GPT-Synopsys to operate Synopsys EDA tools in chip workflows. OpenAI pays a training subscription; both sides share revenue when customers use it. Customer data stays out of training; engineers still sign off. Synopsys also lifted its fiscal 2027 growth outlook to about 15%. This is an announcement, not mass rollout.
Status note: Checked 1 October 2026 against Synopsys press materials dated 30 September 2026 and Reuters reporting from the same investor summit. The partnership is announced with early customer engagements; GPT-Synopsys is not described as generally available to every semiconductor shop. Revenue-sharing mechanics depend on customer adoption not yet disclosed in full.
What the two companies said they are building
The headline product name is GPT-Synopsys. In Synopsys and OpenAI's telling, it is frontier intelligence tuned to operate Synopsys electronic design automation, or EDA, inside real chip-design workflows. EDA is the software layer that turns circuit descriptions and architecture choices into layouts a factory can manufacture. Without that stack, modern processors, accelerators, and phone chips do not move from whiteboard to wafer at scale.
OpenAI will host the model on its infrastructure. Synopsys said it will plug into Synopsys.ai and Autopilot, the company's existing push to automate parts of verification and implementation. The firms also said they are in early technical engagements with leading semiconductor customers, language that signals pilot work rather than a blanket license drop.
Synopsys trades on Nasdaq under SNPS. OpenAI brings the model; Synopsys brings the tools engineers already use and the customer relationships around them. That pairing is why Wall Street treated the news as both an AI story and a Synopsys growth story on the same day.
How chip design got here
For years the industry complaint was time, not talent alone. A new GPU class or custom AI accelerator can take many months of iteration across floorplanning, timing closure, power, and physical verification. Teams explore trade-offs manually or with partial automation. When demand for compute spikes, as it has under the AI build-out, shaving weeks off a design cycle is worth real money.
Synopsys and its EDA peers sit in the middle of that clock. Chipmakers and "fabless" designers pay for licenses and support because the tools encode decades of manufacturing rules. OpenAI's pitch, echoed by co-founder and president Greg Brockman in announcement materials, is to use advanced AI to improve the systems that power AI: let the model navigate more design branches so human engineers can focus on judgment calls the software still cannot own.
That backstory matters for reading Tuesday's news. This is not OpenAI suddenly opening a foundry. It is an attempt to embed a language-and-agent style model inside an existing commercial toolchain.
Money, training, and go-to-market
The deal is structured as partnership economics, not a one-off license fee. OpenAI will license Synopsys EDA for training the model and pay Synopsys a training subscription fee, chief executive Sassine Ghazi told Reuters on the sidelines of the company's investor summit. When Synopsys customers use the product, the two companies will share revenue based on how well the model improves a chip design, Ghazi said.
He added that the agreement was built so it would not cannibalize Synopsys' core business, framing GPT-Synopsys as upside when customers get more value. Reuters noted Synopsys shares rose as much as 7% after the partnership and forecast news, with LSEG data showing the new fiscal 2027 growth view above analyst estimates near 11.19%.
Go-to-market details beyond "early engagements" were not spelled out in the public release. Readers should expect tiered rollouts, NDAs, and customer-specific benchmarks before anyone claims industry-wide speedups.
Guardrails, data, and who still signs off
Chip design is not a chatbot demo. A wrong via or a missed timing path can mean a respin costing millions. Synopsys stressed that customer data will not be used to train the model, that data will be encrypted, and that engineer review remains required. Ghazi pointed to guardrails and traditional sign-off, including physics verification, as non-negotiable steps even when Autopilot-style automation is in the loop.
Brockman, in the company's video message, said the model should work through trade-offs and optimizations during design and help shave weeks or months from the process while bringing more chips to market. That is an engineering claim the market will test design by design. Until customers publish verified cycle-time results, treat the time savings as intent, not a universal benchmark.
Investor day context
The OpenAI tie landed on the same stage as Synopsys' broader financial story. At the 30 September 2026 investor summit, Ghazi also raised the company's fiscal 2027 revenue growth outlook to about 15%, above the analyst consensus Reuters cited. Synopsys separately outlined multi-year targets and fiscal 2027 revenue guidance with a midpoint near $11.15 billion in its investor-day materials.
Those numbers are forward-looking and can move with macro demand, export controls, and customer capex. They still explain why traders bundled the AI partnership with a higher growth narrative in one session.
Where things stand
On the public record from 30 September 2026, OpenAI and Synopsys committed to a multi-year partnership to build GPT-Synopsys, a model meant to operate Synopsys EDA within chip-design workflows on OpenAI-hosted infrastructure, integrated with Synopsys.ai and Autopilot, with early customer engagements underway. OpenAI pays for training access; both companies share revenue tied to customer outcomes. Customer training data is excluded; human engineers retain sign-off duties. Synopsys lifted its fiscal 2027 growth outlook to roughly 15% at the same event.
What is not settled is timing for general availability, pricing for end users, and independent proof that designs finish faster at scale. The announcement is a commercial and technical bet on agentic EDA, not proof that every leading fab has already switched over.
Sources: Synopsys and OpenAI announcement of GPT-Synopsys, 30 September 2026; Reuters reporting on the deal, revenue sharing, and fiscal 2027 outlook at the Synopsys investor summit.