AI in EDA: Practical advice, no matter where you are on the journey

“What’s your advice for getting started with AI in EDA?”

This was a genuine question that someone asked me recently at our User2User Europe event in Munich.

I was there at the event, standing in front of our interactive experience ready to talk about any of our 30+ AI demos when someone asked what I’m guessing many engineers are actually thinking but not all are brave enough to ask. It’s a very real question and I’m so glad they asked it. And it’s one that’s just as relevant whether you’re just starting to explore AI in EDA or already getting deeper into deployment.

My answer in the moment was honest but incomplete: be curious, talk to your product team, explore the demos to identify the potential impact on your workflow. Good advice, but not nearly enough. Behind the question lies multiple layers that deserve a more complete answer, especially as generative and agentic AI capabilities have become a very real part of the EDA conversation.

The moment stayed with me, and, on my return, I sat down to gather my thoughts. What follows is a combination of my own reflections and perspectives from colleagues across the business for anyone wondering exactly the same thing.

The question is completely valid

AI is moving fast; EDA workflows are complex, and the potential cost implications of getting it wrong are real. It’s a change of mindset and workflow that requires significantly more consideration than adopting a consumer AI tool. And while some teams are already deep into AI adoption, running advanced flows and measuring real results, many others are still figuring out where to begin. Both are completely valid places to be.

Questions occur at multiple levels regardless of company size: practical “How will this disrupt my existing flow?”, organizational “What will be the impact for my team?” and commercial “Where does AI actually add value vs. hype?”.

Tip #1: Start with your biggest problem

Start because you have a problem to solve, not because everyone else seems to be doing it. Upgrading software because a new version is available is one thing. Upgrading because a specific limitation is costing your team time and design quality is an entirely different and far more purposeful decision.

Before adopting new AI technologies in EDA, think about the biggest pain points for your team. Where do your engineers lose the most time? Where do iterations spiral into unmanageable delays impacting your tape-out? Where are your most experienced and best minds spending time on something that could be automated?

“We build our AI roadmap around real customer problems, not technology trends. The teams that get the most value from AI start with a specific pain point and that’s exactly where we start too when we decide what to build next.”

Sathish Balasubramanian, Head of Products, Solido Custom IC and Fuse EDA AI portfolios

Tip #2: Test and prove the impact

The temptation is real to deploy AI across all your tools, all your teams, and all at once. But just as taking the time to understand the problem matters, so does resisting the urge to transform everything at once.

Pick one flow, one team, one metric and win there first. In the process of doing so you will have built internal AI champions, created internal playbooks for the practical implications for AI adoption, and built proof points along the way for further AI investments.

Tip #3: Talk to and learn from your EDA vendor

You don’t have to figure it all out alone. Many companies think of their EDA vendor purely as a software supplier. But when it comes to AI adoption, they’re closer to a strategic infrastructure partner.

EDA vendors have AI roadmaps, demos, early-access programs and application engineers ready to help and share their experience. They have deep customer experience and know where AI is delivering results. Also, take a closer look at your existing tools: you may already be using AI-powered features without realizing it. And there are likely new ones ready and waiting in tools you’re already licensed for.

“Once you really understand what a customer is trying to solve, and I mean really understand it, the right approach for their AI deployment tends to emerge. Then it’s about walking that road with them and making sure they meet the goals they set out to.”

Mike Sheinin, Senior Applications Engineering Manager, Solido Design Environment

Tip #4: Measure the impact and share your findings

We all know engineers love to test new things, but AI adoption without metrics is just experimentation. Just as you define design requirements before building, apply the same discipline to AI adoption.

Is there a runtime you want to reduce? Is there a particular task that your team is spending way too much time on? How will the time saved impact your ability to take on more designs?

Even imperfect metrics can create momentum for AI adoption across teams. Sharing findings means that others don’t start from scratch and makes it about adopting a new AI culture, and not just a specific tool.

Tip #5: It’s a journey, not an immediate switch

Think about your personal AI journey: chances are that you didn’t become an AI power user overnight. You tested some tools, assessed the results, the tools evolved and got better, you tested some more, talked to your friends about it, experimented some more, and it’s not over yet…

AI in EDA tools is no different. The learning curve is real, and adoption is a process.

“The best thing I can tell any engineering team is just try it. And once our customers get started, they find they begin to see things differently. The only way to understand what AI can do for you is to get started.”

Jeff Dyck, VP of R&D, Solido Custom IC and Fuse EDA AI portfolios

Back to the question: “What’s your advice for getting started with AI in EDA?”

Every AI journey starts somewhere different. For some it’s a specific pain point, for others it’s a conversation at an event or a demo that sparks an idea.

Wherever you are on that AI journey, you don’t have to figure it all out alone. Reach out, attend an event, explore a resource or request a conversation. We’ll meet you where you are.