Hi, everyone. Good afternoon. Thank you for joining us at the 2023 Needham Technology & Media Conference. My name is Charles Shi, I'm the covering analyst of EDA industry at Needham. It is my pleasure to host this fireside chat with the Cadence Design Systems. Joining me today from the company are Mr. John Wall, our Chief Financial Officer, and Mr. Richard Gu, Vice President of Investor Relations. On behalf of Needham, I wanna thank John, Richard for joining us today. For those who are listening live on this webcast, you can type in your questions in the Q and A box and, or you can just send me an email at cshi@needhamco.com. I'll try my best and work your questions into the flow of this fireside chat.
Before we begin today, the company has asked me to make you aware of their safe harbor statement. The discussion today will contain forward-looking statements and will make use of certain non-GAAP financial measures. Please refer to Cadence most recent 10-K, 10-Q, and the website for discussion of risk factors and their use of non-GAAP financial measures. Without further ado, let's get started. John, Cadence reported a good quarter a few weeks ago, right?
Yeah.
The immediate market reaction was not quite positive, right? I remember.
Yeah.
Although you did recover on the following days. From time to time, I think it. You're familiar with this, right? We encounter situations like this. I mean, the immediate reaction was not so great, but gradually people realized, okay, maybe it's not so bad. Before we really start our chat today, what do you think the market missed or maybe have read too much into your Q1 results? Just wanna give this opportunity for you to clarify a few things.
Thanks, Charles. Yeah, honestly, I can never predict what the market's going to do. Even when I know what the results is, that we're going to announce, I couldn't tell you what the market does the next day. Essentially what we announced in Q1 was, you know, a beat in the quarter, and then we raised the year. I think that got lost in the backlog number 'cause our bookings were light in Q1. Now, the bookings were light in Q1 because we don't have any expiring contracts in the first half of the year. We are. Generally, we have 55%-60% of our revenue coming from the top 40 accounts.
You know, you only have to get to like 150, 200 accounts before you kind of have a Pareto 80% of our revenue, you know, coming from that, you know, group of accounts. We have individual strategies on the top 500 accounts. We have concentration, but we try to make the concentration work for us. Within that, most of our customers do three year baseline contracts, and then they'll come back throughout that three years, and they'll purchase add-ons maybe anywhere from 4-7 times during the three years. We just don't have a lot of contracts, a lot of baseline contracts that expire in the first half of the year. Our backlog declined in Q1.
I think it was a reaction to that. In a recurring revenue model, we're less focused on the size of total backlog. We're focused on growing the annuity or the annual value of that backlog. I think there might have been a knee-jerk reaction to that, mainly among the folks that are more short-term focused. 'Cause Cadence is, it's a compounder really. We always say we're farmers, not hunters. I mean, over a long time horizon, you'll see the benefit of Cadence. You know, we're quite resilient because most of our customers, Our revenue comes from their spend in R&D. We're quite resilient through all of the economic cycles. It's really a compounder in nature, so you have to look over the long term. Hopefully some people saw the opportunity 'cause it looks like the stock price has recovered a little bit since. Hopefully some people saw the opportunity to get in.
Thank you, John. I think, let's talk about some of the recent trends that came up a little bit frequently in my conversation with the investors. Q1 was very strong, a very strong hardware quarter, as you said, on our, on the earnings call. For those who aren't familiar with Cadence or EDA in general, I mean, Cadence is a software company, but they do provide hardware products. In, basically in EDA, hardware means emulation and prototyping hardware that kind of allows chip designers to verify chip designs, and more importantly, I mean, enable customers to really write software codes before the actual silicon is produced. Well, that's a little bit of background for those who are not familiar with the hardware we're talking about.
How should we think about what has been driving the hardware growth in recent years for Cadence? At the same time, why is it that such a, I mean, it's a lumpy business as you characterize, right, for Cadence. From your point of view as a CFO, how do you think about managing that lumpiness as, I mean, lumpiness something that Wall Street apparently really hates?
Yeah. Fair enough. When you say lumpiness, I mean, you mean that that's a part of our business where we get revenue all at once when we deliver. It's like upfront revenue. I mean, the vast majority of our revenue is recurring in nature. I think 85% of our, of our revenue is, you know, recurring, so you've pretty much got the same amount or very similar amounts every week from a, from a revenue perspective on the recurring side. There's 15% of our revenue where, like if you ship a hardware system, once you've installed the hardware system, you get a lot of revenue. The demand for our hardware has been just through the roof for probably 6-8 quarters now. We couldn't keep up with production.
Now we finally expanded production to a level that I think will allow us to catch up with the demand. From the beginning of Q1, we ramped up production, and deliberately did that to address a long backlogger of hardware orders. Hardware order lead times have more than doubled over the course of last year. We wanted to address that. We have customers waiting a long time. They're waiting like six to seven months for our hardware systems to be delivered. Which is not. I mean, that's not ideal. Certainly not ideal when you're trying to just tell the visitor, it's like, "I'll sell you something now, but you can't have it for seven months." We thought, well, we had to ramp up production.
The key thing is when we ramp up production, we'll produce the same amount of hardware, substantially the same amount of hardware this quarter as we did last quarter. We ramped up production. We intend to keep it at those levels. There's probably four main constituents for that hardware that comes off the production line. The biggest one are that queue of customers that have purchased the systems outright, and they're waiting for delivery and installation. Internally, we have a cloud infrastructure, a hardware cloud infrastructure that we provide for customers, and we have to put those hardware systems into that cloud infrastructure to provide that. That's typically, I mean, some of this, some of our emulation hardware is priced like a Ferrari.
There's a lot of customers that can't afford or might not have the requirement to own a Ferrari, but they might want it for the weekend or a month or something like that. We direct them to our cloud infrastructure, but that's been maxed out for a long time. We'd already put some systems into that. We have, I mean, our own internal teams use our hardware for their own verification when they're testing things. Also then we do put demonstration models out to customers to demonstrate the machine, particularly in new accounts. We're proliferating the hardware into new accounts.
With those, customers might use them for one or two quarters, and then we'll convert those into a sale when they see the value of those hardware systems. In Q1, because the length, the queue was so big, we prioritized customers first. I remember we had that discussion internally, and they said, "Well, do you want us to spread out the deliveries to customers across the quarters so that you don't have so much lumpiness in Q1 versus Q2?" We said, "No, we always put the customer first." I always tell our team that, "Keep the, you know, put the customer first. We'll explain everything.
Mm-hmm.
In Q1, pretty much every system went to customers that were on that backlog. We had a lot of upfront revenue in Q1. In Q2, we'll produce the same number of machines. Excuse me. More of those systems now in Q2 will go into new accounts where we believe we'll sell hardware in the second half of the year, but we have to provide demo machines now to show them the value of those hardware machines. In the past, we would have used a combination of demo machines for bigger opportunities on-site and the cloud infrastructure. In some cases, we demonstrated using the hardware in the cloud, and then you can own the system if you purchase it.
But we've been so maxed out on the cloud, and we haven't had enough systems to use for demo machines. Some of the production in Q2 is going to that. With some of the production going to that, naturally, we'll have less hardware revenue in Q2, which of course will be described as lumpiness in terms of our hardware. But demand is really strong. The reason it's strong is, I mean, clearly you see the growth and complexity of design. You hear Adam talking about, you know, that's our customers are, I mean, the most advanced ones are able to fit 100 billion transistors on a 1-inch-by-1-inch chip. That's inevitably, we believe, going to 1 trillion.
You know, 900 billion additional transistors over time, that's complex for anyone. For simulation, the simulation challenge or the emulation challenge with something like that is two to the power of 900 billion. It's more complex for that group than anyone else. I think there's a secular demand for those systems. It's also I think we've been quite disruptive in, l ike, there was, l ike, there's typically a chronological order to our customers, you know, use our tools for chip design, verification, packaging it forward, and system analysis. In that chip design and verification, even in simulation, you might, s imulation is so complex that our customers will use our full suite of tools.
They'll use Jasper for formal simulation, like Axiem, software tools to try and narrow down the areas they want to look at. On the emulation side, the real value that I'm told that our customers get from that is the production cycles for our customers have shortened so much. I mean, they're trying to release new products every year. To do that, when you produce a chip design, you need to have an emulation system so that the software people can develop the software apps that will go on that chip. Now you can use a prototyping system for that because the chip would be hard. In that case, the chip would be hardened. In an emulation system, you have the advantage of tweaking the chip because the chip isn't fully hardened.
In some of our customers, the software folks, they will ask the chip designers to tweak their design to optimize the software. It allows the teams and our customers to collaborate more than they were ever able to before. Those systems become very, very popular. It does, like you say, it results in some lumpy revenue from time to time based on where the hardware that comes off the production line is dedicated. Like I say, in Q1, we had such a long lead time.
I would expect to get back to more normal lead times by the middle of the year, which I think is important because it sets us up for a stronger second half if we have demo machines in place and we have that capacity in the cloud infrastructure to show how good these systems are to drive more sales in the second half.
Got it. Got it. I think it sounds like it's fair to say, you are maintaining the same supply level in Q2, and it's just a sum of the revenue opportunity for part of the shipment going out in Q2. It's a little bit deferred in the future quarters. It shows up as if the software revenue is coming down a bit in Q2, but it's actually just a little bit deferred in terms of when the revenue gonna show up. Maybe, John, a follow-up question here.
As some investors are kinda wondering if is China part of the reason for the hardware strength in recent quarters, and if yes, are you worried some of the China strength, may be a result of the stockpiling, for the fear of potential, U.S., further U.S. export controls?
So hardware, we do sell a lot of hardware into the China region. I mean, our hardware is popular all across the globe. The ratio of hardware to software revenue for Cadence is higher in China than it is in other regions. Normally, I mean, it's not always the case, but when you do have a strong hardware quarter, you probably have a strong China quarter. It's not always the case because I mean, it just depends on where the hardware is getting delivered in any one quarter. That's why we tell people not to focus too intensely on any single quarter.
I mean, internally, we always look at a three year time horizon because that's our typical baseline contract with these customers, and we're always trying to grow the annual value that we're extracting. But, the timing of a shipment, I mean, just the timing of a hardware installation, if you're going out to deliver a hardware system in December around the holiday time, and you might find, oh, well, there's no one around. The customer is not signing off on it until January. But, that could be the difference between hardware revenue in one year versus the next. It doesn't really make a huge amount of difference to us because we take a longer-term view on it, on everything. Essentially, yeah, I think in China that, you know, they love our hardware as much as anyone else.
I don't think there's any pull forward going on there. I think just the timing of deliveries. If anything, the concentration in Q1 was us ramping up production capacity so that we could catch up, and there just was a heavy proportion of China customers in the queue. I think, again, a strong hardware quarter would typically have a strong quarter for China. like if I look at China over the last, I think, through the last three years, 15% of our revenue came from China. in Q1, 17% of our revenue came from China. I would expect it is closer to 15% for the year, given that two of the last three years were 15%. You know, maybe it's a little bit higher.
With all that extra hardware, maybe the, maybe China grows this year slightly better than average, in which case it might tick up a little bit higher than 15+. Our, our customers, the hardware is popular right across the globe. There's, you know, I don't use any pull forward as anything.
Got it. Got it. Thank you for that. John, the other metrics I think investors look at when they look at Cadence are your backlog as well as the implied bookings number, right? I think you sort of touched upon this a bit. Your backlog is down by a smidge this quarter, I mean, this past quarter, to $5.4 billion from $5.8 billion a quarter ago. It is still nicely up from, let's say, three years ago when you kind of reported $3.7 billion for March 2020, right? March 2020, every time you mention March 2020, everybody start to real think about, okay, that's COVID and is that a little bit random why I pick this comparison.
Yeah.
I think because you often talk about the business on this three-year on a three-year basis, I'm asking this accordingly. Now my question really is this, is the backlog reduction in March 2023 quarter, is it temporary, or is it just one of those lumpy metrics, like the hardware sales that's kind of inherent in the EDA business? How should we think about when every time you report a quarter, then people look at the backlog, oh my god, backlog is down or booking number is down, but how should people think about that?
Right.
Yeah.
That's a good question, Charles. Again, I wouldn't focus too much on the total backlog. For us, it's, you know, the total backlog doesn't matter so much as the annual value of that backlog to us. Now, in terms of Q1, I think you're being kind calling it a smidge. I thought it was significant to drop by $400 million. But a big portion of that $400 million is us delivering the hardware systems, catching up with some hardware systems that were in backlog. But the main driver, the majority of it is because of, we didn't have a lot of contracts expiring in Q1 or Q2.
You know, when we're more back-end loaded or second-half loaded for contract expirations this year, like if you have large contracts that expire in Q3, well, at the end of last year, so there's nine months in backlog for those. At the end of Q1, there's six months in that you declined. At the end of Q2, you'll only have three months in backlog. Of course, by the end of Q3 you'll have renewed it. When you renew it, of course, they'll renew for, typically for three years, you have 12 quarters, and it's just a big jump. First half, I would expect us to burn a little backlog. Most of it happened in Q1.
By the time we get to the second half, we have a lot of expiring contracts that come up for renewal, and we'll expect that to recover going forward from the end of Q2. Really it's just the timing of when, like I say, it's just the timing of when contracts naturally expire that's causing the overall total backlog to decline. The annual value of the backlog is continuing to increase. Like, when I look at our recurring revenue that, you know, every time we do renewals, typically customers will we have 99.9 something in this, in terms of renewal rates with customers. The only customers that don't renew either have expired or they've been acquired.
Practically everybody renews, and it is by far the exception if someone renews for a lower annual value than we had previously. I mean, every year you'd expect growth. When someone's renewing a three year baseline contract, now typically they'll have add-ons throughout the contract, it's exceptionally rare for anyone to want to renew for or to have less of a requirement. That's someone that's on their way towards expiration probably. Practically everybody renews, and practically everybody renews for a higher price than previously. Again, when you look at total backlog, that's the function of timing. Total backlog is the annual value of your backlog multiplied by how much time you've signed customers up for. For us, the annual value metric is most important.
Hey, Charles, also just a quick point. I think as we highlighted on earnings call, right, the underlying underpinning tailwinds for the design activities remain very robust for us now, so and that hasn't changed.
That's very true, yeah.
Yeah.
Thank you very much. Maybe let me expand a little bit on this back half loaded contract renewals. You also touched about that it's very rare for customers to want to sign a smaller contract when they renew, when they do renewal. You also mentioned about your retention rate, right? Is very high, right? I mean, if it's not 100%, it's very high in the 90s. Now I think one of the questions we've got is given where the macro environment is and given where the semiconductor cycle is, given where the all the tech companies or a lot of the system companies you are servicing today, they're putting more cost discipline to that. Could you possibly see a relatively, I mean, smaller growth in terms of the contract value when they do renew in the second half of the year? I think that this is the question that people got a little bit, let's say, skeptical about so-called second half recovery. I think a lot of companies have been talking about that, the investors probably just naturally a little bit skeptical about that. Can you address that? Y eah.
Sure. Sure. Just so I understand the question, are you saying it's clearly the environment this year is a bit more subdued than it has been in more recent years, and it's not a good time to be doing renewals, I think is what I hear you saying?
Yeah. Yeah. The question is it a not a good time to sign up bigger contracts when they—
Yeah.
Need to renew? Yeah.
Okay. Look, it's a great point 'cause I wanna tie that back into the previous question that you had because the reason, you know, you may ask me like, why do we have so few renewals in the first half of this year? Or why do we have so few contracts expiring in Q1 and Q2 of 2023? I do think it's not a coincidence that it's three years from Q1 and Q2 2020 when the pandemic broke out.
Back in, for those that have been following us for a long time, we always have our management team off-site and I mean, in the past that I would be challenged to come up with something for the team as a scenario and let's get the team working together on different team-building exercises or projects and stuff like that. In August of 2019, the yield curve had inverted. I went to the management off-site and I told them this, "Okay, yield curve is inverted. If I go back over the last 50 years, it suggests that when the yield curve inverts, there could be a recession in April 2020." Pure fluke, right? Pure fluke. I asked them for the exercise. I thought it was an interesting exercise for us to do.
I told them that, "Okay, you know, it might not predict a recession." We certainly weren't predicting COVID. I asked them that if we knew that a recession was coming in April 2020, what would you do differently? I wanted them all to think about the business and Cadence and what they would do differently. There was a bunch of discussions about that topic about that all contracts that are coming up for renewal in Q1 and Q2, maybe I wouldn't want to be negotiating a three year contract if in the middle of a recession. That would, you know, people are gonna be more tight about their dollars. That, you know, that any softness you're kinda extrapolating out for multiple years if you try to do it then.
What they looked at were the more critical or higher value renewals, they wanted to try and close them early. We made a project out of it and said, "Okay, let's try to do that." What we agreed amongst ourselves was, let's not give any value away if we're doing something early, though. That if any customer wanted additional value or a discount or something to renew their contract early, we'd say, "Well, no. We don't wanna do that." We just wanna take some business off the table, set us up for the next few years. Of course, what happens?
We had done that, when you get to Q1 and Q2, when you've already taken some of the business off the table, in Q1 and Q2 of 2020, when people are very nervous about signing up to three year deals, our sales guys are parked on the. We're focused on growing that annual value. To ask a customer to sign up for three years when they're so uncertain, is difficult. Many of the contracts that came up, even though typically people do baseline renewals for three years, our sales guys just said, "Okay, if you're not comfortable with the three year value, why not just extend it by one year, extend by two years on your existing deal?
We'll come back and talk about this next year, focus on getting through the pandemic. The result of that is the knock-on effect is that by the time you come around to when if we were doing three year renewals then when they would come around to naturally be expiring, there's just not a lot of contracts expiring. That's the phenomenon you're seeing in backlog. Doesn't hurt us. In fact, I think it helps us that we're not having to negotiate in more difficult environments. We probably got lucky twice, got lucky in 2020 and getting lucky again in the first half of 2023. I think what typically happens. Again, we're focused on the annual value. We want to continue to grow the annual value.
If there's kinda softness in the general market, we don't have to do a three year arrangement. That it can be one year or two years until the customer is comfortable, more comfortable to sign up for better terms for us. What I do find is when I look at the, say, the weighted average duration of bookings at Cadence, if I look over any three year time horizon, the weighted average duration is typically somewhere between 2.4-2.6 years. It typically averages about two and a half. The reason for that is most customers over a three year time horizon will do a three year baseline contract. Some will do shorter, some will do longer, but the majority are doing three years.
Because customers are purchasing add-ons throughout that three-year contract, what we do with add-ons is that if you, if you don't have access to enough tools or enough licenses or some technology you didn't purchase in your original baseline contract and you want to add it on, we'll do an additional or an addendum to the contract for you. If you're like a year into your three-year baseline contract, the add-on will be for two years and because they all co-terminate with the, with the baseline contract. As a result, because all the add-ons are naturally less than three years, as a result, you typically average about 2.5 years over a 12-quarter window. The, in any one quarter, like in, like a pandemic quarter, it's going to be less.
That's not, that's not a reflection of the entire business. That's I do think over three years, typically things seem to even out.
Got it. Thanks, John, for the great transparency here. I think we really appreciate that color like this. Certainly it sounds like a backlog and bookings. Well, you wanna be flexible in terms of when to renew the contract, when, what kind of contract to discuss with the customers, whether you want to focus on the annual value that you already have and including add-ons and onto the so hopefully at the, when the time is, it becomes better, you can sign a bigger contracts with the customers. Maybe I wanna touch upon a little bit more on the IP side of the business.
I know it's a, probably a smaller proportion of the Cadence business compared to your closest peer. We understand you don't want to chase revenue in IP. You want to grow profitably. I think the Cadence did set a goal of growing IP business at a roughly low teens % CAGR a few years back. When I look at the IP growth numbers, so the year-on-year numbers, it feels like you were growing at 10%-ish year-on-year in 2021, 2022. Is 2023 on track to grow above 10% so that 'cause I kinda think that you do need a little bit above, well above 10% to really get to that low teen % CAGR, if you would think about the three-year horizon when we're, when we calculate the CAGR?
That's a, that's a fair question. Very interesting observation there, actually. I would, again, encourage you to look over a longer time horizon. The answer to your question, the short answer to your question is yes, we'd expect to grow faster than that. If I look, like if I go back to 2018, our IP business, our IP revenue was about in and around $250 million. That's 2019 we grew that to $300 million. Now 2020, you'd expect, okay, 250, 300 should be 350, right? Of course, it's a 53-week year. We had that funny thing going on and but I think that ended up closer to about $370. Then you kind of normalize, you get back to 2021, about $400.
2022, I think it was about $440. I think that's where you get into the 10%, $400 to about $440. For 2023, I mean, if we've gone essentially $250-$300, I know $370 to in the with the extra week and everything to $400, $400-$440, I want like $490-$500, right? I mean, you would expect something like that. I mean, if you go through that $440 in 2022 again, if you look over a three year time horizon against the $300 in 2019, that's about 13%-14% annual growth. That's low teens. I think $440 to anywhere near $500 for this year, again, is low teens.
I think the business is delivering low teens, but the most important thing to us has been we've asked them to try and deliver low teens, but with increasing profitability year after year. That's been our focus. That's, when we review contracts back around the 2017, 2018 timeframe, we found that we grew really fast up to that point in IP, but sometimes we were signing up to revenue growth. It was like, it was just so expensive, and it just wasn't working. We were tying up like, precious engineering resources on activities that were not generating income.
I always tell people, "What do you care more about, your gross pay or your take-home pay?" I think everyone would kind of raise an eyebrow if you said, "Look, I can double your gross pay, but your take-home pay will be less." You wouldn't want that. We tell people that, look, we wanna run this. I mean, people will tell you that I'm like a broken record at Cadence. Most meetings I say, "Okay, would you do if this was your family business, is this the decision you'd make?
Would you sign up to this contract?" We always think about it because as a, when you're thinking of yourselves as a compounder, we have an amazing strategy that we're becoming more and more important to the, to society and our customers in terms of like we're the glue that holds everything together. You know, there's such a high dependency on Cadence that, it's important to think long term. Like I say, compounder nature, we think long term. We, we try to think, well, what would you do if it was your family business? We, we talk to investors that way. I mean, we take our responsibility, and treat it accordingly. That's it.
It's really important to us that we feel we're managing investors' money, that we want to invest it for the highest return. I view returns in, on an income basis. That's why the focus has been like that. There is some business that's available every year within IP where you can sign up for any revenue growth you want if you don't care about how much it costs you to deliver it. I don't wanna do that because that's not sustainable long term.
Got it. Got it. Thank you. I think I wanna move on a bit to discuss with you on the PNL lines. I think this is a, this is a question always a little bit worth checking in with you, John. I think you famously set no operating margin target, but you do have a flow-through target, right, 50%, meaning 50% of the incremental revenue falls directly to the operating profit line. A lot of investors kind of interpreted that as Cadence operating margin probably will converge to this 50% figure eventually, right? I mean, mathematically, right, the path to that 50% may be a little bit like a asymptotic curve , right?
I know we always set a high bar for Cadence. Is there anything structural that can help you move that 50% flow-through target higher? Maybe, I mean, asymptotically, Cadence operating margin can actually be tracking to somewhere above 50% over the long term. I know we set a high bar, but I wanna hear your thoughts.
No, great question, Charles. But yeah, I don't often have people ask me a question on asymptotic basis. Essentially, what you're saying is like, when I say, like, it's like that question you would have had in school that, oh, you know, John has to travel a mile. He covers half the distance in 10 minutes. You know, how long does it if he covers half the distance every 10 minutes, how long does it take for him to get to reach the mile? Of course, you never reach it, because if you're only ever going half the distance, half the distance, you never actually get to your goal. I think that's what your point is with the 50%, right?
Right.
Okay. The, well, okay, one thing is, although we kind of set expectations with people that we're trying to achieve 50% incremental margin, the one thing we discover with Cadence is it's, it scales dramatically well, particularly if you focus on your core center of gravity. I mean, we always felt that our core center of gravity is that we were destined to be the largest core EDA company. That's because, you know, where our roots were in analog, and we felt analog to digital is like orchestra to rock band, and it was easier for orchestra players to play in rock band. Naturally, we felt it would only be a matter of time before we get to 50/50 in digital. We would retain the majority of the market in analog.
If your destiny is the largest company in core EDA-We've built our strategy around that because then, you know, we look at what our customers use our tools for. It's chip design, chip verification, then they do packaging and board. We have a packaging and board business that, you know, then they go into system analysis. We're branching out into system analysis. We're building that kind of unified platform, if you like. We're targeting areas where we feel we have the right to win and where we can benefit synergistically because it's a close adjacency. What we found is that Cadence is an, EDA is a remarkable business, but it scales really well. The challenge in the past was how do you grow it?
I think we have a really great strategy and a path for growth that, I mean you can see phones have become smartphones, cars are becoming smart cars, buildings becoming smart buildings, cities will become smart cities. It's just, it's pervasive and we're in the sweet spot for all of that. I mean, prior to phones becoming smartphones, there was like a social media kind of explosion that doesn't benefit us at all. All these other things benefit us and I think it's really, really helpful for us. When you think of what I looked at then was, okay, at the time that I stepped into the CFO role, Cadence had got up into the mid-20s for operating margin, and our incremental margin was only I think 11% at the time.
I thought, we're investing in the wrong areas. We're not investing in the areas where you have the greatest opportunity for growth. I thought, the challenge at the time was revenue growth was decelerating. I think we went through a series of years where revenue growth went from like 12% to 11% to 10%. It was like heading down towards 7% or 8%. I felt we had to turn that around. One of the issues was investing in the wrong areas meant that, look, while we were improving operating margin in the 2012-2016 periods, we were improving by not investing in certain areas that was impacting the pace of new product introduction. Of course, that was having the knock-on impact of reducing revenue growth.
What was really important was to look at Cadence on a Rule of 40 basis, add your operating margin to your revenue growth, and see if you could turn it around by allocating your investment more intelligently. When we did that, we found that we were able to turn around the pace of new product introduction. We also found that I set a goal of 50% plus incremental margin. I wanted every dollar of revenue growth to drop at least $0.50 to the operating income line. This is for non-GAAP, 42 for the GAAP because we spend about 8% of revenue on share-based comp. We've been strict about that as well, trying to hold that.
On a non-GAAP basis, we try to drop $0.50 for every $1. Over the course of what is it? Over that last six, seven years, we have ranged 53%-56% in any one year. 53%-56% over a rolling three-year period in incremental margin, 52%-58% in any single year. We've beaten that 50% in every one of the last six years. That's six years, very different years. That's two contract cycles. Had a pandemic in the middle of it. You know, if you're traveling at that pace for six years in a row, I see no reason why you can't continue to do that.
Of course, but if let's say we reach a static 56%, you will have that asymptotic issue that maybe you get close to 56%, but you never get over it. I think the 50%, we've been beating 50% so consistently, I think that will help. What you'll find though is that, you know, the pace of acceleration on margin probably slows the closer you get to your incremental margin. What we thought was really important to us is make sure we have an efficient machine. We're doing things, we're scaling as well as we possibly can.
Then you look to expand into other areas with investment, so that you can put more and more revenue through this machine that's spinning off amazing cash flow and amazing profits for investors. That's been our focus. That's what we've been trying to do. Every year we're investing more and more in new product introduction. What's interesting, back at the time when I started this, I just did like a straw poll of all the engineers that were running the R&D groups. They were spending probably one-third. My simple question was, how much of your time is being spent on products that generate revenue today? How much of your time is spent on products that have not been released yet?
Only about a third of their time is being spent on products that hadn't been released yet. I felt that, if two-thirds of your time is spent on maintenance of existing products, that's not gonna help you with new product introduction. I think that's pretty much flipped. I mean, in the six or seven-year period, as much as we're spending on R&D these days, the majority of that R&D spend is not contributing a dime to revenue this year. It's all new products being generated for the future. It's us building out our capability. As long as we keep the machine operating as efficiently as we can, it should spit out, you know, operating income and cash flow for the benefit of investors for years to come. That's why I say we're farmers, not hunters.
We plant now, harvest later.
Got it. Got it. I think that's a very good way for us to really wrap up today's discussion. I think, John, you provided a lot of transparency, especially the transparency to how as a CFO of Cadence, you think about managing the business from a CFO point of view. Really appreciate that. I think we are about to have to wrap up the session today. Really thank you, John and Richard for joining us today. Thanks everyone on the line and hope everyone has a good rest of the day. Good. Thanks, Charles. Thanks, Richard.