He believes that reuse is most beneficial when there is reuse, and outlines how to teach designs to AI and adapt it to new designs through self-learning. He noted that chip designs are largely derivative, so reuse helps the AI learning process.
The introduction of AI in the design community is inevitable, he said, but Andersen argued that this could be a positive, just as some machines replaced some manual labor in the first industrial revolution.
“You look back and say, Who would want to do this very hard work? Especially when it comes to the physical work. … If you use those tools correctly, you actually raise the level of work. And the same thing happens with chip design. If I offered you a technology to automate generations of things like RTL and timing constraints, or to fix someone’s misspellings or errors in a script, I think most people would say, “It’s great because I’m more productive.” The good thing is that designers can spend more time doing creative work. While AI is good at automating tasks, coming up with completely new ideas is still the domain of humans, he said.
“[AI] is replacing the task, but it is not directly replacing the engineer because the engineer can do more. And the other thing is, it’s not like a million integrated circuits are being created, so fewer people are needed to do it. It’s quite the opposite. We basically don’t have enough people to design these chips. In fact, there are many predictions from various consulting firms that up to 30% more work will be required for the shortage. From that perspective, I think everyone is very welcoming of it because it really addresses the requirements that are being brought to the chip industry. ”
Synopsys will host SNUG Silicon on March 11-12, 2026 at the Santa Clara Convention Center, California.
Electronics Weekly Podcast CHIIPS
