Anirudh Devgan: Yes, Vivek, what I would say qualify AI is another way of providing dramatic automation, that’s — and this history of our industry is to provide more and more automation. And I would say AI is like third wave of massive automation and productivity improvement. I mean, the first wave was moving up abstraction level going from transistor level to gate level to RTL level to C level. That’s one wave that drove our industry, which is — the second wave in the last 10 years is parallelism. Cadence has invested heavily in massive parallelism whether it’s on-prem or on the cloud, and starting from 2013, we have a lot of paddle products. And then, now, the third wave of productivity and automation is using AI or what I always call AI for optimization now called reinforcement learning or generative AI.
So, we have a history over the last 20, 30 years of doing this kind of — and in all cases, it doesn’t cannibalize. Actually, the activity increases and amount of software used and amount of optimization that happens, more and more designs are done with the available resource headcount. So I expect AI to do that. I mean, in general, the benefit is so huge that when you get these new capabilities, you use it to do much more efficient design. Okay. I’ll give you an example, starting to — some very senior people in our industry, and one comment has been that the Moore’s Law has slowed in terms of actual improvements you’re getting from one node to another node, when you went from 65 to 40-nanometer, you get X amount of improvement, and now when you go to 5 to 3, that used to be 20%, now is — they went to 50%, now maybe it’s like 10% improvement.
And even our tools like Cerebrus or a lot of the other AI-based tools can provide easily 5% to 10% improvement in PPA. So you are getting improvement from better algorithms, which are similar to half or full node of a process technology improvement, okay? And that’s just huge. The amount of improvement you can get from this third wave of automation using generative AI is huge. So — and I think that’s one metric that we have published a lot in terms of how much PPA and product improvements we are getting, and that’s what we closely monitor with our customers. And I think — and the other thing is to apply that across other things, not just digital implementation, to apply to package design, to board design, to simulation. And what I think it — opportunity it gives us is, over time, we can get higher share of R&D invested in automation, because the complexity of these things is going up exponentially, so the headcount can’t go up exponentially.
So, the headcount, I expect, still will grow five, 10 years from now in terms of R&D spend in our customers, but we can get a bigger share of that R&D spend applied to automation.
Vivek Arya: Thanks, Anirudh.
Operator: Your next question comes from the line of Gianmarco Conti with Deutsche Bank. Your line is now open.
Gianmarco Conti: Hi, there. Hi, Anirudh and John. Thanks for taking my questions. So, on my first one, I just want to touch base again on the AI suite piece. Firstly, whether you’re entering a more mature phase of the price discovery mode as you’re accelerating volumes? And secondly, perhaps thinking about the specifics of what are the AI demand drivers here, whether those are correlated to increased design at the lower nodes, are there other trends that we should be aware of other than the increase in complexity, i.e. some specific trends in the industry that you’re seeing beyond the semi players, particularly as you’ve mentioned, of course, optimality in SD&A? Is there anything that we can sort of like track or understand what’s going on behind the bonnet? I’ll ask a follow-up question after. Thank you.
Anirudh Devgan: Absolutely. I think, we talked about chip, right, and how AI can help this third wave of automation. But the same thing can be applied at package and system levels. I do want to highlight that. We have talked for a few years on importance of merging system and semi and how system companies are doing semi and then we can also provide solutions for system design and analysis like thermal simulation, electromagnetic simulation, fluid dynamics. And traditionally, those areas were — first of all, the simulation capabilities were — could be improved, just the raw simulation speed. So, for example, Clarity for electromagnetic, we got order of magnitude improvement versus traditional methods, because applying our computational software expertise, we can do a lot more simulation.
But on top of that, there are two other things that can jump to our system design and analysis, and you’re seeing that even our growth that you’re seeing. So, one thing is use of GPU acceleration. And we talked about this in my prepared remarks. And I think GPU acceleration is significant for system design and analysis there. And the other thing is applying AI on top of simulation. So, EDA has a long history of optimization, not just simulation. An AI for optimization of generative AI is really new for system design and analysis, because they are barely — even the simulation capabilities were not keeping up, but they didn’t really have that much optimization capability. So, the response we are getting from Allegro X AI and Optimality is huge, because not only simulation is possible for the first time, but the optimization of simulation.
Because in general, if you’re doing some thermal design or design of the data center, yes, you want to stimulate of a car, but you want to optimize, whether it’s the shape of the wing or placement of the racks in the data centers. And all this is really now possible with generative AI. So, I think the impact on this SD&A will be profound apart from the impact on chip design. And we are the company that can combine those two things and apply AI to both of these areas.
Gianmarco Conti: Great. Thank you. And just my second one is on M&A, whether there’s any M&A in sight or plan on increasing the portfolio perhaps in SD&A — sorry, in systems and analysis, or is 2023 consistent with the plan of distributing cash to shareholders via buybacks?
Anirudh Devgan: Yes. I mean, in general, we want to start with the strategy, and we feel we have a very, very strong strategy with intelligent system design, this combination of silicon system and data. And we are very pleased with our progress, and we continue to grow organically and perform well both in terms of revenue growth and margin. So that’s our base strategy, base outlook. Now, we always evaluate M&A as it comes up and if it’s a good return for our shareholders and good return in terms of R&D. But in general, we are pleased with our strategy and our organic execution.
Gianmarco Conti: Got it. Thank you.
Operator: Your next question comes from the line of Blair Abernethy with Rosenblatt Securities. Your line is now open.
