Anirudh Devgan: Yes. One thing to add to what John said, I think you know this already, but one thing to emphasize, like typically, when people move from one node to another, right, so we are at 5-nanometer, a lot of the design’s at 3-nanometer, then 2-nanometer, 1.4-nanometer, 1-nanometer. So, we have like 10 years of node migration ahead of us. So, any time you go from one node to the next node, the number of transistors effectively doubles in the chip. So, number of blocks effectively doubles and complexity doubles. So, for the same chip being done, it has more things in it so that normally requires more use of our software. So, that’s one thing just to emphasize to our investors is that whenever there is a node transition, and this is happening in the past, but this will continue to happen for next 10 years.
So, that also requires more hardware capacity. That requires more simulators. That requires more place and route runs to be done. And of course, with AI, we can make that even more productive. So €“ and Moore’s Law is not slowing down for at least 10 years, okay. And on top of that, you add 3D-IC. That adds another 10 years of. So, we have a lot of sustained growth that the customers will build amazing products, and they need more and more of our software and hardware to do that.
Ruben Roy: Yes. Thanks Anirudh. I guess that’s where I was going with the question. I mean obviously, with Cerberus going €“ running on top of Innovus. And I am just kind of trying to think about how €“ it seems, and you guys have talked about this, greater portion of R&D budgets from both semiconductor and systems companies just going into kind of the suite of tools that you offer. So, anyway, thank you very much, and congrats on a great quarter.
Operator: Our final question will come from Andrew DeGasperi from Berenberg. Please go ahead. Your line is open.
Andrew DeGasperi: Thanks for taking my question. I guess first in terms of the analytics products that you mentioned, I was just wondering in terms of your customers, how steep of a learning curve is it for them to adopt us? And what is the kind of timeline that it typically takes from when you introduce it to them to when they are kind of using it in terms of their day-to-day processes?
Anirudh Devgan: Yes, that’s a very good point. So, I think it’s something like Cerberus AI, our use model is very similar to what they are doing right now, okay. So, we don’t change any of the interfaces. And if they are used to running Innovus, then Cerberus will run on top of it, almost like a cockpit, but it will do those experiments that they were doing manually, they will do it in an automatic way. But the interfaces are similar. The commands they are used to running are similar. But instead of manually running and trying different things, Cerberus will do that automatically. So, typically, we have seen €“ that’s why you have seen a lot of kind of uptick to Cerberus and these AI-based tools because they don’t fundamentally require a new working model.
It just takes away some of the mundane tasks that the customers were doing. And similar things are true for optimality, similar things are true for Verisium. There is always some learning gap, but it is, our customers are very smart users anyway, right. They are designing all these complex chips. So, this is nothing that they can’t pick up in a pretty quickly.
Andrew DeGasperi: Yes. That’s helpful. And then one question for John, in terms of the share repurchases, I think you bought 1 billion or so last year. This year, you are guiding for a bit below that. I was just wondering, what was the thinking behind that? Is it a function of investing in the R&D in terms of the different €“ the three different themes you mentioned earlier, or is it potentially for additional M&A that you are planning for the year?
