Chip design is one of the world's most complex engineering tasks. A single modern processor consists of billions of transistors, and placing them flawlessly on a small silicon wafer takes months, sometimes years. On September 30, a major partnership claiming to radically accelerate this process was announced: chip-design software giant Synopsys and OpenAI are working together on a joint model that creates chips with the help of artificial intelligence.
The agreement was announced at Synopsys's investor summit. According to Reuters, the parties work on a revenue-sharing model: OpenAI pays a training subscription to learn from Synopsys tools, and when customers use the finished product, revenue is split based on how much the model improved the chip design.
The essence of the partnership
The model, named GPT-Synopsys, covers the full chip-design cycle: from describing chip schematics in a code-like language to placing billions of transistors on a silicon wafer. The OpenAI model learns to use Synopsys software tools and helps engineers consider design trade-offs and optimization options.
OpenAI co-founder Greg Brockman explained the goal in the announcement video thus: to save "weeks, months" from the design process and "deliver more chips to the world."
What does the model learn?
Per Reuters, GPT-Synopsys will not be a traditional language model but a system adapted to a specialized task. It learns two things at once: first, the language for describing chip schematics — engineers express an idea in a code-like format, and the model turns it into a physical design; second, using Synopsys's own tools — i.e., instead of "guessing" the finished solution, it works through existing software like a professional engineer.
The difference of this approach is that the model doesn't draw the final answer itself — as an "assistant" working alongside the engineer, it evaluates design trade-offs (e.g., the balance between speed and power consumption) and suggests options. Final approval always stays with classic tools.
Why did Synopsys agree?
At first glance a question may arise: won't the AI model make Synopsys's core business — design tools — unnecessary? Company CEO Sassine Ghazi rejected this concern in an interview with Reuters:
"We structured the deal so it doesn't eat our business. Because we're giving the customer more value, this will be incremental growth for our business."
Ghazi stresses that the model's work is double-checked by Synopsys tools based on traditional compute methods: "The model needs these guardrails to check physics. Validation with the highest level of reliability — what we call 'sign-off' or 'ground truth' — is very important." In other words, AI proposes, and the final decision is confirmed by classic tools operating on the laws of physics.
Market reaction
On the day the partnership was announced, Synopsys raised its fiscal 2027 revenue growth forecast to 15% — notably above the 11.19% analysts expected per LSEG data. After the announcement, the company's shares rose up to 7%. This shows market confidence in the prospects of AI-assisted chip design.
Why does this matter now?
The artificial intelligence boom itself has intensified the chip shortage: the compute needed to train frontier models is rising sharply. Shortening the design cycle is not just engineering convenience but a factor affecting the growth pace of the entire AI industry. OpenAI's entry into the upper part of the chip chain — the design stage — shows the company aims to cover not just model supply but the whole technology stack.
This deal is also a lesson for software companies: even in the most closed and specialized markets, AI enters as a "partner" — not as a competitor. Synopsys's revenue-sharing model could become a template for other engineering-software makers.
The Uzbek context
Uzbekistan is not yet a notable player in the global semiconductor chain, so the direct economic impact of this news is limited. But there are three indirect signals.
First, engineering education: as AI-assisted design tools cheapen and spread, hardware engineering may "democratize" like software — which in the future means remote highly-skilled jobs for Uzbek engineers.
Second, a strategic takeaway: even closed, high-skill fields like EDA aren't staying outside AI's influence. So workforce training programs should put AI working skills at the center.
Third, a signal for investors: capital flow around chips and AI infrastructure continues, keeping the question of technological sovereignty on the agenda.
