Artificial intelligence is moving beyond software development and into one of the most demanding areas of modern engineering: semiconductor design.
OpenAI and Synopsys have joined forces to develop a specialised AI model called GPT-Synopsys, designed to help engineers build, optimise and verify the chips that power smartphones, cloud computing and advanced AI systems.
The idea goes beyond using AI to answer technical questions or write snippets of code. The model is being developed to work directly with Synopsys' electronic design automation (EDA) software, interpret engineering results and test different approaches to solving complex design problems.
The companies announced their multi-year partnership on 30 September 2026. Under the agreement, OpenAI will license Synopsys' design tools to help develop the model. The two companies will also collaborate on research, product development and commercial distribution, with a revenue-sharing arrangement. Early engagements with leading semiconductor customers are already underway, according to the companies. <Cite refs={["turn864840search0","turn864840news6"]}/>
The timing makes sense. As AI systems become more capable, demand for advanced chips continues to grow. Designing those chips, however, is a complicated process involving billions of transistors, strict power limits and extensive testing before a design can move into manufacturing.
OpenAI and Synopsys want to bring AI into that process, giving engineers more ways to explore design options and reducing the time spent on repetitive tasks.
The goal is ambitious, but there is an important distinction: GPT-Synopsys is being developed to automate parts of chip engineering and support human designers, not to eliminate the need for engineering expertise or independent verification.
WHAT IS GPT-SYNOPSYS?
GPT-Synopsys is a specialised AI model being developed to work with Synopsys' semiconductor design software.
General-purpose AI assistants can explain technical concepts, generate code and help developers investigate problems. Designing a semiconductor requires considerably more. Engineers must turn specifications into working circuits, manage complex electrical and timing requirements, and make sure the final design can be manufactured and operate reliably.
EDA software helps them carry out this work through circuit design, simulation, performance analysis and verification.
The proposed GPT-Synopsys system is intended to make AI a more active part of these workflows. Instead of simply recommending a solution, it could operate the relevant tools, examine their output and use the results to decide what to try next.
Under the companies' plan, engineers could assign specific objectives to AI agents. These agents would run design tools, review the results, make adjustments and repeat the process until they reach a suitable outcome or require further engineering input. <Cite refs={["turn864840search2","turn864840search5"]}/>
That would represent a meaningful shift in how AI is used in engineering. The model would not just produce an answer; it would participate in a process where its decisions can be tested against technical requirements.
From instructions to repeated design improvements
Consider an engineer who needs to improve a chip's performance without exceeding a defined power budget.
An AI agent could explore different configurations, run the appropriate Synopsys tools and compare the results. If one option misses a timing target or consumes too much power, the system could use that feedback to explore another approach.
The process could continue until the design meets the required targets or an engineer needs to step in.
How well this works will depend on the model's ability to make sound decisions, interpret tool output and handle complicated engineering tasks. The partnership provides a framework for developing those capabilities, but real-world results will determine how useful the technology becomes.
WHY CHIP DESIGN IS BECOMING A MAJOR AI FRONTIER
There is an interesting connection at the heart of this partnership: the AI industry needs increasingly powerful chips, but designing those chips is becoming more challenging.
Modern processors contain enormous numbers of transistors and interconnected components. Engineers must balance performance, electricity consumption, heat, physical size, manufacturing requirements and cost.
These factors are closely linked. A change that improves computing speed might increase power consumption. Reducing the physical size of a circuit can make timing requirements harder to meet. A design that performs well in an early simulation might still fail later verification checks.
Finding the right balance takes repeated testing and specialised knowledge.
Meanwhile, cloud providers are expanding their AI infrastructure, and semiconductor companies are developing processors for increasingly demanding workloads. Faster design cycles could help them respond to this demand, although manufacturing capacity and supply-chain limitations would remain.
This is where the OpenAI–Synopsys partnership could make a difference. If AI can explore more alternatives and handle parts of the design process automatically, engineers may be able to develop complex chips more efficiently. <Cite refs={["turn864840search0","turn864840news6"]}/>
The benefits could extend beyond major technology companies. Smaller semiconductor businesses might also gain access to more capable design assistance without having to expand their engineering teams for every specialised task.
They would still need qualified engineers, suitable computing resources, licensed software and access to manufacturing and verification services. AI could help make the work more efficient, but it would not remove those requirements.
POWER, PERFORMANCE AND AREA: THE BIG THREE
A central goal of semiconductor design is balancing three measures known as PPA: power, performance and area.
Power: How much electricity a chip consumes.
Performance: How quickly it completes its intended computing tasks.
Area: How much physical silicon space the design occupies.
Engineers cannot always improve all three at once. A design that excels in one area may create compromises elsewhere.
For example, a processor built for a large AI data centre may need high computing throughput, but power efficiency is also important because electricity and cooling costs add up quickly. A chip designed for a mobile device may place greater emphasis on battery life, heat management and a compact physical footprint.
AI-assisted design could give engineers a way to explore more possible configurations, compare their results and identify promising alternatives.
The partnership also targets timing optimisation and verification closure. These stages help establish whether a design meets its timing requirements and passes the necessary technical checks before manufacturing.
The real test will be whether GPT-Synopsys can improve these outcomes without introducing errors or creating additional work for verification teams. A design that looks better on one measure is not necessarily the best overall solution.
WHY HUMAN OVERSIGHT WILL STILL MATTER
As AI takes on more engineering tasks, reliability becomes a central concern.
A semiconductor design cannot be accepted simply because an AI model produces a convincing explanation or suggests a plausible modification. It must meet strict technical requirements, and mistakes discovered late in development can be expensive.
According to Reuters, Synopsys CEO Sassine Ghazi said the model's work would continue to be checked using the company's conventional engineering tools. These tools provide established methods for assessing whether a design meets the required specifications, offering an important check on AI-generated changes. <Cite refs={["turn864840news6","turn864840search1"]}/>
That distinction matters. AI may suggest a change, but engineers still need evidence that the resulting design works as intended.
Models can misinterpret information, make incorrect decisions or recommend changes that appear reasonable but fail under more demanding conditions. Independent verification remains essential.
Verification is necessary, but it is not a complete guarantee
Conventional engineering checks provide important safeguards, but they cannot automatically eliminate every risk.
The system's reliability will depend on how well it handles complex requirements, how thoroughly its output is tested and whether engineers can identify problems that automated checks miss.
Security is another concern. Semiconductor designs often contain valuable intellectual property built up through years of research and substantial investment.
OpenAI and Synopsys have said the system will include enterprise-grade security, governance and access controls. They have also said customer data will be encrypted in transit and at rest and will not be used to train the model. Configurable data-retention, auditing and permission controls are also part of the stated approach. <Cite refs={["turn864840search0","turn864840search2"]}/>
These protections will matter to companies considering the use of AI with proprietary chip designs. Customers will still need to review the actual deployment arrangements, data-handling terms and access permissions before sharing sensitive information.
A NEW COMMERCIAL MODEL FOR CHIP ENGINEERING
The agreement is about more than developing a new AI model. It also creates a commercial framework for combining AI with specialised engineering software.
Reuters reported that OpenAI will pay Synopsys a subscription fee for access to its tools during model training. Once the product reaches customers, the companies plan to share revenue based on its effectiveness in improving chip designs. <Cite refs={["turn864840news6","turn864840search1"]}/>
The arrangement gives both companies a reason to make the technology useful in real engineering environments.
For OpenAI, the partnership offers an opportunity to expand into a specialised industrial market. It also supports the development of AI systems that can carry out complex, tool-based tasks rather than relying mainly on conversational interactions.
For Synopsys, the deal could increase the value of its existing software. If AI helps engineers become more productive and complete additional design work, demand for advanced engineering tools could grow.
There is, however, a commercial hurdle to clear. Customers will want evidence that the technology saves time, improves design quality or reduces costs without introducing unacceptable risks.
The revenue-sharing arrangement creates a path towards commercial adoption, but it does not guarantee success. GPT-Synopsys is still under development, and its long-term prospects will depend on performance, pricing and customer confidence.
HOW THE PARTNERSHIP COULD CHANGE ENGINEERS' WORK
Semiconductor engineers spend considerable time investigating failures, adjusting parameters, running simulations and reviewing results across different stages of development.
AI agents that can manage some of these tasks could reduce repetitive work and give engineers more time to focus on difficult technical decisions.
They might be able to test more design alternatives within the same development window, investigate problems more quickly and compare possible solutions with less manual effort.
But chip design involves much more than meeting a single performance target. Engineers must also account for manufacturing yield, reliability, security, operating conditions and the commercial requirements of the finished product.
AI could help evaluate these trade-offs, but engineers would remain central to defining specifications, reviewing results and approving designs.
The more likely near-term outcome is a shift towards engineers working alongside increasingly capable AI systems, rather than fully autonomous chip development.
Tasks involving routine tool operation, repeated optimisation and initial problem investigation could see the greatest changes. At the same time, skills in system architecture, verification, AI-assisted engineering and technical analysis may become increasingly valuable.
The effect on employment is less certain. It will depend on how reliable and affordable the technology becomes, how widely companies adopt it and which tasks can genuinely be automated.
THE WIDER RACE TO BUILD BETTER AI CHIPS
The OpenAI–Synopsys partnership reflects a broader cycle taking shape across the technology industry.
AI needs powerful chips. Developing those chips is difficult, so companies are exploring ways to use AI to speed up the design process. If those tools help produce more efficient processors, the resulting hardware could support further advances in AI.
That creates opportunities for businesses working across AI software, semiconductor design and cloud computing.
It also highlights the strategic importance of electronic design automation software. EDA tools help turn a chip's initial specifications into a design that can eventually be manufactured. Improvements at this stage could influence how quickly new semiconductor products move through development.
Still, designing a chip faster does not mean it can be manufactured faster automatically.
Fabrication capacity, advanced packaging, testing equipment and access to leading manufacturing processes remain important constraints. Even substantial improvements in design productivity would need to translate into manufacturable products at competitive costs before their broader impact becomes clear.
EARLY CUSTOMER TRIALS WILL PROVIDE THE FIRST REAL TEST
OpenAI and Synopsys have said that early technology engagements with leading semiconductor customers are already underway. However, the cited announcements do not establish a broad commercial launch date or provide independently verified performance results for GPT-Synopsys. <Cite refs={["turn864840search0","turn864840search5"]}/>
Those early engagements will help show whether the model can handle real engineering problems reliably, beyond controlled demonstrations.
Several questions remain:
How much time can GPT-Synopsys save on actual chip-design projects?
Can it improve power, performance and area without creating new problems?
How much human supervision will it require?
Will smaller semiconductor companies be able to afford it?
How effectively will it protect proprietary design information?
Can it work across different chip architectures and complex engineering workflows?
The answers will help determine whether GPT-Synopsys becomes a widely adopted engineering platform or remains a specialised tool for selected applications.
The multi-year agreement gives both companies time to develop the technology. Ultimately, adoption will depend on demonstrated results rather than the promise of AI alone.
FACT CHECK METER
FACT METER: PARTNERSHIP ANNOUNCED; TECHNICAL AND COMMERCIAL RESULTS STILL TO BE ESTABLISHED
Status: OpenAI and Synopsys announced their multi-year partnership on 30 September 2026. Reuters also reported on the commercial arrangement and the continued role of conventional verification tools. <Cite refs={["turn864840search0","turn864840news6"]}/>
Confirmed details:
OpenAI and Synopsys are developing a specialised AI model called GPT-Synopsys.
The model is intended to work with Synopsys EDA tools for semiconductor engineering.
The agreement includes software licensing, joint research and development, commercial collaboration and revenue sharing.
The project targets areas including PPA optimisation, timing and verification workflows.
The companies have outlined security protections for customer data, including a commitment that customer data will not be used to train the model.
Early technology engagements with semiconductor customers are underway.
Important distinctions:
The model is under development, and broad commercial availability has not been established in the cited reporting.
Potential time savings and design improvements remain outcomes to be demonstrated, not guaranteed results.
Conventional engineering verification will remain necessary.
The agreement does not establish that AI will replace semiconductor engineers.
AWX VERDICT: The partnership marks a concrete move towards AI-assisted semiconductor engineering. Its significance lies in giving AI a more active role inside professional chip-design tools. Whether it delivers lasting value will depend on measurable productivity gains, dependable results and the ability to meet the security and verification standards required for semiconductor manufacturing.
WHAT TO WATCH NEXT
Customer trial results: Look for measurable improvements in design time, quality and engineering productivity.
Commercial availability: Updates on launch timing, customer access and pricing will help clarify the product's market potential.
Verification performance: The ability to produce designs that pass conventional engineering checks will be a critical measure of success.
Revenue contribution: Future financial disclosures may show whether the partnership becomes a meaningful source of income for either company.
Data security and governance: Customers will want clear information about intellectual-property protection, permissions and data retention.
Industry adoption: Watch whether other semiconductor companies expand their use of AI agents for design, simulation, verification and optimisation.
THE AWX BOTTOM LINE
OpenAI and Synopsys are bringing AI into one of the semiconductor industry's most technically demanding tasks: designing the chips behind modern computing.
GPT-Synopsys is being developed to do more than generate suggestions. It is intended to operate engineering software, interpret results and refine designs against demanding technical requirements.
If the approach works as intended, engineers could explore more alternatives and complete parts of the development process more efficiently. But the technology will have to prove itself in real projects, and rigorous verification will remain essential before designs move towards manufacturing.
The partnership also points to a broader change in the AI industry. Companies are increasingly looking beyond chatbots and code generation towards systems that can work directly with the professional tools used to build complex products.
AI may not replace chip designers, but it could change how they design the next generation of semiconductors.
ATTLAS WIRE | AWX WORLD NEWS
Real Stories. Lasting Impact.
Comments
No comments yet.