I think we're live. Welcome back, everyone. I hope you enjoyed lunch. Just before media earnings, which is great because it wouldn't be as interesting, everyone being on their phones. Today we have the pleasure of having John Wall, Chief Financial Officer of Cadence Design Systems, and Richard Gu, Master Head of Investor Relations. Before we start, I'm going to read a quick safe harbor. Today's discussion will contain forward-looking statements, including Cadence's outlook on future business and operating results. Due to risks and uncertainties, actual results may differ materially from those projected or implied in today's discussion. Great. Now that we got that out of the way, perhaps let's begin with setting the tone for the EDA landscape.
I'd love to hear what you're seeing in the market right now that is exciting Cadence the most as an opportunity to expand more of its portfolio in, especially given you have such a close relationship with your core customers like NVIDIA and the likes. It'll be great to hear more about that.
Thank you, Jenny. Thank you for having us here as well. That's really appreciated. I guess the most exciting thing at the minute has to be like AI supercycle, right? I mean, our customers are pushing the boundaries of design and chip design. We see them pushing those boundaries with things like 3D IC and the full system simulation, advanced packaging. It's not really all about one company either, in terms of one partner. We benefit from having close partnerships with leading customers like NVIDIA, but also like Intel Corporation, Samsung, TSMC, and a whole host of others. We're privileged to have the customer base we have. I think it's the breadth, having access to the breadth of those customers and partnerships and the position we are at a time when there's just huge opportunities right across the landscape, whether it's semi-companies or hyperscalers or system companies.
I think that's probably the most exciting thing about where Cadence sits right now.
Yeah, that's fair. It's a pretty diversified landscape you have there across all. That's really good. Given that we're talking about AI and the AI supercycle, I think it's good to spend a few words on perhaps the Cadence AI portfolio, which seems to be expanding both in scope and in customer base. We'd love to hear a few words on when you expect, firstly, to see this materially contributing to revenues, and secondly, whether you're seeing continuous demand beyond the top five customer base. I think it was well understood by the market that at the beginning, when Cerebrus was launched, it was sort of like penetrated at the beginning among the top five customers, right? It's obviously used for the most advanced and leading edge nodes and those types of designs. It'd be good to hear where is it positioned now?
Great question. Certainly, that's how we began. I think we're well beyond the top five customers now. I think there's broad proliferation across most of our customer base. It's still probably early days for monetization because we've largely a ratable revenue model. I mean, 80% of the revenue is ratable. The vast majority of that is daily revenue on a subscription basis. As we proliferate into these accounts, it tends to create an uplift, but then that kind of flows through over time. It's exciting to see the adoption across the board. Typically, it takes us a couple of contract cycles to fully proliferate. In the first contract cycle, there's a lot of proliferation of the technology and use for the technology.
What we're seeing now, kind of we're halfway through that kind of two contract cycles, is that we're starting to renew accounts, renew baseline contracts that had AI in them last time. We're seeing increased adoption, increased usage, and then the license counts are starting to increase significantly on some of those contracts. Like I say, what we're seeing across the board as they adopt more AI tools is that not only are they achieving productivity benefits in their own design cycle, but faster time to market. Faster time to market results in earlier revenue recognition for what it is that they're trying to release. There's immense value being extracted from the use of those tools by our customers. I think we're at a point in the cycle now where they're willing to share some of that value with us. We're seeing that through increased license adoption.
Yeah, because that's a good point. I do remember that for every license of Cerebrus, there was sort of like an increase, like a natural increase in innovative, right? It's like a scaling effect. You know, if you buy more of those Cerebrus licenses, then you have to buy more of the legacy licenses. That's kind of what's driving right now in that growth bucket.
Absolutely, yeah. I mean, one of the beautiful things about Cadence, I mean, we're very diversified. We have multiple lines of business. We did an analysis a long time ago just to understand the profitability profile of all the businesses. Cadence started as an analog franchise, really. I mean, the Virtuoso team there has been tremendous for us for decades. When you look at Cadence, of course, it's no surprise that the most profitable business at Cadence is our software business. In software, you could bifurcate our software business into two groups. Software tools where one license needs one driver, like a Virtuoso license, is one engineer uses one license of Virtuoso. If you had a 100-person analog design house, they're probably buying 100 licenses of Virtuoso from us. They won't buy 110 until they hire 10 more engineers. Same with Innovus, typically, but on the digital side.
What we had with Cerebrus is that one engineer controlling a Cerebrus cockpit has the ability to use 10 licenses of Innovus. Not only are you selling Cerebrus as an AI tool, but it's pulling through an extra nine licenses of Innovus. I think you've seen that happen as well in terms of the whole, you know, how the stack works. Do you want to talk about any of that?
Sure. Yeah, John. We have this fantastic platform called Cadence AI, on which there are like five flagship products, with Cerebrus being the tip of the spear. We actually recently launched Cerebrus AI Studio, which is truly a game changer. It's the industry's first multi-user and multi-block, you know, agentic AI products. What it does is it's allowing one engineer to parallel run multiple kind of designs concurrently, dramatically enhancing the sort of like the time to market. You can reduce the time to market by 5x- 10x by doing that. In the meantime, it increases the improves the PPA benefit by 10%- 20%. I think we talked about customers like SD Micro and Samsung using the technology and really benefiting and harvesting the tremendous improvement. I think we're deep in that process of launching various agentic AI products to help our customers on their AI journey.
Pretty impressive.
Yeah. I guess beyond Cadence AI, I think the word agentic has been thrown out in the air more than Crypto in 2021. I'm just curious to hear what you think about whether you see a real path for agents redefining the actual design workflow, i.e., providing an opportunity like a customer such as, I don't know, NVIDIA, Broadcom, Qualcomm, whoever, to actually cut a single-step process, such as synthesis or placement route, or if it is fictitious in a way, or something that would happen maybe 10 years down the line. You hear Synopsys saying about that, touting the word agentic. I'm curious to hear your thoughts about it.
You sound skeptical, Jenny.
I don't know.
Agentic AI is real. I mean, our Jedi platform is an agentic AI platform, and that's allowing our customers to use agentic AI for things like verification, test optimization, and optimization loops and test generation. They'll use it for things like that. I mean, the synthesis and full placement routes, that might be a longer way off, of course, right?
Yeah.
There are some benefits that customers are getting, and they're real benefits. They're real productivity benefits. It's allowing them to cut design costs and get more done faster. I mean, would you agree?
Yeah, absolutely. I think in the back end, we have a lot of, obviously, agents for implementation, for verification. There are lots of opportunities in the front end, too, in terms of how do you leverage the large language models to translate the language to RTL code and the test bench, things like that. Our engineering team are hard at work, you know, trying to make sure we capture fully that opportunity to help our customers on the journey.
Yeah, there's a long runway here. I mean, this is still only early days. I think the benefits that we've seen, even in the short term, are real.
That's fair. That's fair. Maybe just like talking about system designs, given the recent closing of, I don't know how to ask this question, but it's major news, right? Given the recent closing of the Synopsys merger, I think it would be helpful for investors to understand whether user chatter around an expected material increase of pricing or their licenses could ultimately benefit Cadence, in that there would be at least initial opportunities for exploration of alternatives. Secondly, maybe even a natural pull from any integration or support issues. I guess where I'm heading at is, given the incredible demand we're seeing in anything simulation these days, is this deal net negative or benefit for Cadence?
No, look, I don't see it being a negative for us. I think it's neutral to positive. I got to give kudos to Synopsys. I mean, they're a great competitor. You can't have the Yankees without the Red Sox, right?
Exactly.
The competition between the two companies drives us both to be better. I don't think Cadence would be as good as it is today, or Synopsys would be as good as it is today without the competition we have between each other. I do think in the last five, six years, we have made significant advances. I think the boost that they'll get from adding Ansys will make them more competitive again against us. There are huge benefits that you get from putting everything together. I mean, chip design, verification, packaging board, right through to system analysis. If you can do all of that in one platform, that adds a lot of value to customers. I can see how they're likely to deliver more value as a combined offering. Maybe that results in higher prices.
Whenever there's a change like this in the market, it's always a catalyst for customers and incumbents to review the current incumbents against what other alternatives are out there.
Exactly.
I can see that being positive for us over the longer term. I don't see any downside, really.
I kind of agree. I think there's also an element of customers are really dying to see that shift left of simulation prior to everything else. I think digital twin had never seen, I mean, it hasn't seen innovation in so long. For them to be able to have that concurrently when they design their systems is such a clear, obvious, hence why there was excitement around this deal, particularly on the method side of Synopsys. It'd be interesting to see where the market, the CFD, MFA market will be heading in about 10 years from now.
Yeah.
OK, cool. Shifting maybe a little bit into financials, John. Given you delivered an outstanding Q2 print, like it was fantastic and surprising quite a few of us, especially on China, could you share some color, and how should we be thinking about backlog development for both 2025 and 2026? Particularly, I'm going to press on this point, any detail on upcoming renewals that allows us to shape our model better?
Yeah, sure. OK.
We know Q3 is a big renewal, like renewing quarter.
Big question. Let me unpack it a little bit.
Yeah.
In terms of backlog, we finished last year with record backlogs of $6.8 billion. Now we've eaten some of that backlog in the first half of the year, where I think we closed at $6.4 billion at the half. We're seeing a strong booking environment in the second half that we're confident that we'll finish this year with new record backlogs. It will be above the $6.8 billion. That's based on strong renewal activity, but also strong add-on activity from customers. I think we're seeing strength, broad-based strength right across the board in all lines of businesses at Cadence. At Cadence, we're privileged that we have five businesses under the umbrella of Cadence. Typically, there's always one that's dragging their feet a little bit.
At this moment in time, it just feels like everything is delivering right across all geographies and across all businesses. That bodes well, and it's very, very positive for backlog for the remainder of this year and heading into next year. Also, I know there's a lot of focus on, you know, in some businesses, there's a lot of focus on the timing of renewals with specific customers. We're blessed to have such a broad diversity of large customers that you really can't design an electronic product without using Cadence. We've such a diversified group of customers that we're not dependent on any one renewal in any one quarter. I think we have strong renewal activity over the next or right through the remainder of 2025 and into 2026.
Even though this was, I mean, $6.8 billion was a record backlog, and we should see that be higher by the end of the year. You're setting up yourself for a pretty tough comp to next year. How should we think about it? Particularly because there was a big hardware push this year, obviously from the next generation of emulators and prototypers, and you closed out that air pocket quite impressively. How should investors think about your setup into 2026 when you have a tougher stance on, well, tougher comp on hardware? Although you might have a record backlog by the end of the year, you're still facing that hardware and IP and simulation strong backlog.
I totally understand the question. I guess the nature of printing record after record after record is you give yourself tough comps.
Exactly.
Now, I think one thing that you should make sure you're aware of for this year is that last year you mentioned the air pocket. We launched our new hardware system, Palladium Z3, our new emulation system at the end of March, beginning of April of last year. It kind of created an air pocket for us where pipeline opportunities for our older Z2 system, people kind of waited six or eight weeks because they wanted to see what Z3 looked like and how long it would take to get access to Z3. Many of them still followed through and bought Z2 because they could get access to that quicker.
Yeah.
As a result, this year we had easier comps Q2 over Q2 than we'll have in the second half of the year. The second half of the year, some of the hardware activity that got sucked out of Q2 just kind of got caught up in Q3 and Q4. The second half of last year was very strong, and we're lapping those in the second half of this year. Still very confident, though, that we'll beat those comps. The revenue from our hardware business is really throttled by our production capacity, and we keep raising our production capacity to meet demand. It's hard to keep up with demand because the growth in complexity and design shows no signs of slowing down.
Our verification group, when they're pitching to us for budgets, they'll often say that, look, you know, and we know everyone's chasing Moore's law, and that's tough when they all need investments. If complexity is growing by x, complexity in verification is growing by 2 to the power of x because they're trying to deal with all the variables that you might have in a verification situation. When they're trying to emulate all of that, there's no slowdown in that emulation. The pace of innovation has to keep up, and the customer's requirements, that's why nobody wants to miss their silicon first-time round.
No.
You have to spend money on verification. There's huge demand for it. Nothing's getting less complex. We've proven, I mean, our Palladium Z3 is our showcase product. It's Cadence on Cadence. It's our own custom chip designed using Cadence IP and Cadence tools. There's a wide moat around it. We think we have the two best emulation systems on the planet right now. The second best is Z3, is Z2. That's a strong position to be in. Bookings are quite volatile. Bookings in Q4 will be a lot higher than bookings in Q1. It's just that way every year. People kind of use up the end of their budget for this year. They're probably spending some of their budget, like they have visibility into their budget for next year. A lot of that gets signed and committed in Q4. Q1 is kind of dried by comparison.
We manage that volatility at backlog and in the booking side. From a revenue perspective, revenue is based on the amount we can produce and issue from time to time. We keep increasing our production capacity. We don't want to increase it too much because we like the price points that we're at, we like the availability. The scarcity works for us. We try to manage lead times somewhere between eight weeks and, ideally not as high as 26 weeks. It went as high as 28 weeks before, and we had to really ramp up production to get that back. Somewhere between eight weeks and 20 weeks is probably the sweet spot. It's kind of pushing towards 20 weeks right now.
Yeah.
We are ramping up production in the second half of this year to help us deal with that.
That's good. I think you had mentioned that if someone wants to buy an emulator today from you guys, you won't have it ready by the end of the year, right?
We have systems, right? We have systems, but there's such a backlog of orders. If someone has an immediate requirement, often the desire for a new emulation system is probably triggered by an upcoming project. If you have a project starting in November or December, you're probably looking for an emulation system now. Right now, it's tough to get that, you know, get in the queue. If I just put you in the normal queue for that, you might not get it by the time you need it.
Right.
We'll have to work with you and see if there's a way that we can.
Makes sense.
Meet the deadline.
Maybe shifting a little bit of topics and speaking about renewals, I think everyone has a few key topics on the top of their minds. I believe Intel is one of them. I know we should be cautious of any direct customer mentions. I was hoping you could share a few words on how you view their upcoming Synopsys renewal and how you think about your internal IP development for 18A and 14A. I think the market is mixed about this. Some are thinking there is a clear in to steal market share, given the obvious relationship with Lip-Bu, whilst others and investors are a bit more skeptical and receptive about the shift as a sudden replacement would be seen as radical.
I think there's a sort of like a mix between, oh, you clearly have an in and there's more upside than downside, versus, oh, maybe there is upside only in the incremental portion of spend of Intel. I'd love to hear your thoughts about that.
I'm sure Intel is clearly an opportunity for us. I'd prefer to be more indexed to Intel. It just so happens that we're probably more indexed to the likes of a TSMC than we are to an Intel or a Samsung. They tend to spend more with our competitor than us. I do think that creates more opportunity than downside for us as things change there. We have IP and tools that are all silicon ready for 3 nanometer, 2 nanometer, and beyond and signed off at these foundries. We're in a position to help them as much as they want to help. I think the biggest opportunity there right now, basically, if anyone's going to turn around Intel, it's probably Lip-Bu. Lip-Bu, I've seen him at close quarters. He's an exceptional person, one of the greatest capital allocators I've ever seen.
I think he'll realize that Intel could, I mean, what they're spending on EDA is probably nine times more on people in EDA than tools. I think that ratio of people to tools probably needs to change. If you're going to make changes there, you need to spend more on tools, which might end up being that there's more upside on the tool side for both ourselves and Synopsys, even though Synopsys is very highly indexed there. They could probably do more with less people.
That's interesting. Yeah, that's fair. On the IP side particularly, how indexed are you on the most advanced nodes for the foundry business?
Sorry?
For the foundry business of Intel Corporation, like 18A and 14A, are you continuously updating your IP for those nodes?
Oh, absolutely. Yeah, like I said, we're ready to help as much as they need us. We're still improving at all these nodes.
OK, maybe going back to China, I guess this is how I think about it. The elephant in the room is what do you think about the region in terms of it being a sustained stream of business, or whether this will slowly die out and its customers shift to local vendors? The way I'm thinking about it is that China will still stay a part of Cadence, and customers in the region that cannot access U.S. technologies or are being pressured not to will likely move their operations in other APAC regions. I think that China, for the time being, will likely be a bit more volatile, with revenue growth potentially being flat to declining as a baseline. Am I thinking about this the right way?
I think if there's volatility at a regional level, I don't think you'll see as much volatility at the top level. Our regional revenue is based on consumption, based on where licenses are used. If there's any trend, what we're seeing is that, let's say, maybe five, 10 years ago, you might have done a software arrangement with a customer in China, and 100% of their engineers were based in China, and the licenses were being used in China. These days, that's happening less at the bigger customers. They might still have a large R&D group in China, but they probably have multiple R&D groups around the world, and the licenses are being used in different countries.
What that does to our geographical mix of revenue is it means if the previous deal, if you had 100 licenses all being used in China, and then on the new deal, maybe it's 150 licenses, but 110 are being used in China. China will grow by 10% from 100- 110, and you'll have growth of 40 licenses in other parts of the world. It's a consumption. It's based on consumption. It's very hard for us to predict that. We're much more confident in the top-level revenue line and the growth that we can generate there than the actual geographical mix of that revenue.
That's fair. Yeah. I mean, I was at DAC this year, and I was talking to a few of the folks at Empyrean. It was surprising to me to hear that they're actually not doing the full flow end-to-end EDA software, which means that even, so say as a baseline case, you know, China is not an issue in terms of the geopolitical tensions. If China becomes like it was two, three years ago, it's still, there is virtually no end-to-end competition in China, right, for you guys. The people there are still over-indexed to either Cadence or Synopsys, or potentially Mentor Graphics. Are you seeing any competition heat up from the result of these tensions? As far as I see it, which was only in June, they're not doing physical verification, not doing functional verification, not doing placement route. You're not catering the full needs of a chip designer.
No. I mean, if we're seeing competition, we're seeing competition for headcount in China, in terms of talent for China.
Right.
In terms of competition for bookings opportunities, revenue opportunities with customers, it's not so much. It's taken decades for us to build relationships and close trusted partnerships with our largest customers and to build out these full flows and the sign-offs with all the foundries for those flows. It's very hard for anyone to replicate that in a short space of time. That'll take decades. I just think it's hard for anybody starting off to catch up with Cadence and Synopsys at this point. Not impossible, but we're talking a decade-long journey.
Right. The times are definitely high. I think even the big guys, like if you think about Google with their deep, there's like a deep chip product that it's like a pointal solution, effectively, in EDA, but it's like an implementation pointal solution. It's not really, I mean, yeah, it can help. There were like some studies about some research scientists using it and saying it's purely academic, but it's not fully replacing even the single implementation pointal EDA that Cadence offers.
That's right. We don't see those customers sitting down and having meetings trying to figure out how to replace the optics.
No, exactly.
I mean, they're basically trying to figure out how to leverage our technology to solve other problems.
To solve their problems.
Yeah.
A final one on AI. It would be helpful to also frame the AI landscape in terms of startups coming to play in the space. There are a fair few impressive ones at DAC this year. I'm curious to hear your thoughts on their claims of trying to reshape the EDA landscape and whether you see these as, like, again, to my point, single point tools, threats, but not real competition given to the end-to-end engine that Cadence provides, or whether you're starting to see some real speeds. I'm curious to see both if it's a threat or if it's sort of good for you guys or even M&A.
Yeah, so not a threat for the same reasons. Like I say, it takes trusted partnerships, and the full flow takes decades to produce. We love the fact that there's innovation in EDA again. AI is generating a lot of that. We see the startups as well. There's been a dearth of startups in EDA for the longest time. That's probably because of the size of ourselves and Synopsys. It's really, really difficult to compete. Now that we're seeing some innovation coming through, I think that's really, really positive. We might see some really good, clever point tools. Like you say, it's an opportunity for us to pick up the best of them. I think the best of them will end up as part of the flow at a company like a Cadence or a Synopsys.
That does make sense. Because when you're doing these kind of acquisitions, it's more like talent-wise acquisitions, right?
Yes.
Like you're buying the company because there's maybe 10, 15 of those engineers that are very hard. They're working very hard on a single niche tool and niche solution, which you might be able to leverage in your bigger end-to-end solution.
Absolutely. The character of Cadence is, the company has over 13,000 people now. More than 90% of the company are engineering qualification. Incredible engineering base there. The core nature and DNA of the company is that it's a company created by engineers for engineers. They're always trying to solve their peers' problems for them. We're always trying to help everybody do their jobs better. The nature of those engineers is they always prefer to make rather than buy. Even if an acquisition is out there and there's an acquisition opportunity, just because it's an opportunity doesn't mean we'll go after something like that. Typically, what we look for is, will it further our strategy? Will it accelerate our timeline in terms of the strategy that we're trying to implement? Does it bring new technology and new talent to Cadence that we don't already have?
The most important question is, is it at a price point that makes sense for us? Often the answer to that is no. We'd rather invest in our own team and our own people or hire some people to go after those opportunities. The nature of Cadence is we're long-term in nature. We always say we're farmers, not hunters. We would rather plant now and grow over time than do an acquisition.
That makes sense. I guess on this point, it's interesting because we're living in this year and even in 2024 where the AI talent war is huge, right? We know that on the one hand, electrical engineers, so you know people that go into becoming a verification engineer, that's already a limited supply of those people. The people who are doing AI software development and hardware engineering, it's even more of a niche. I guess the question would be, how are you retaining your AI talent? Especially given the big shift and the big push that you're doing with Cadence.ai, which requires knowledge of both of those things.
Yes, yeah. I mean, great question again. In terms of AI talent, I do think people that are interested in that field are naturally attracted to a place like Cadence because you get the opportunity to work on that. I'm not sure AI is the greatest name for what it is. That probably confuses a lot of people. I mean, Applied Statistics might be a better name for it in terms of more a clearer description of what it actually is. Engineers at Cadence, for years and years, like when Anirudh was looking at our simulation capability and the power of our matrix multipliers, when he started doing that exercise, we realized that maybe 30% of Cadence's revenue was from simulation activity. Just having that, like the capability, the knowledge, the opportunity at Cadence, I do think it tends to attract talent.
We can always engage with our customers, many of our customers that we compete with for that talent, that also rely on us for tools. There's opportunities for them to outsource some of their AI needs to us as well if the talent is tough to come by.
That makes sense. I guess I think we have time for maybe one to two questions, especially because there's NVIDIA earnings coming up.
No one cares about NVIDIA. It's just the biggest company in the world. I guess I'd like to conclude with asking about, you know, to your point, acquisitions. You mentioned you'd rather be a farmer than a hunter. That's been the case for Cadence Design Systems, right? If you look at it over time, you've done very few acquisitions, and the ones that you've done are quite small. I mean, the largest one was Beta CAE Systems, which was sort of like a needed one, right? Because you're now competing against Ansys. They're doing structural analysis. You ought to compete also in that space, and you have the customer. The customer basically is some cross-selling. I guess the question would be, within the M&A landscape, what is your vision for the next two to three years? Are you still going to be doing some small bolt-ons?
If you're doing these small bolt-ons, do you see them being more in IP, simulation, or a combination of both? I don't see it to be in EDA, but maybe I'm wrong.
Right. There's not a lot in EDA to buy.
Exactly.
Ultimately, it's customer-driven. We're not opportunistic. Just because something's available for sale doesn't mean that we're interested. We'll basically take our lead from customers. We're always focused on solving customers' problems. When we look at M&A opportunities, we prepare to make rather than buy. The nature of the M&A we've done has been what we describe as tuck-ins internally. I don't think we've spent more than 1% or 2% of our market cap at any point in time on an acquisition. Naturally, as you're getting bigger, maybe we're doing slightly bigger dollar value acquisitions. It's just based on the size of the company. It's really tuck-in opportunities. The type of thing on the system design analysis side, there's great opportunity. We're very focused on driving profitable and sustainable revenue growth. We're focused on that bottom line, making sure earnings are growing.
We like to cannibalize our own share count by buying it back.
Eat it.
When we're looking at those opportunities, often we're looking at areas where, with limited investment, it opens up new revenue streams for us. If we have a lot of simulation talent and capability internally within Cadence as a core competence, and then we can make a small acquisition, a small tuck-in acquisition that gives us some domain expertise, it's a pretty low-stakes bet. Your very limited investment is opening up a new opportunity. Beta was a great example where they had capability that we never had. Not only did it create opportunities for us with Beta and being able to work with them more closely, it created a whole bunch of pull-through opportunity for the rest of our system design analysis portfolio products. It's been very, very successful so far.
Amazing. John, thank you so much. Richard, too.
Thanks, Jenny.
[crosstalk] Pleasure.
Thank you.
Thanks everybody.