益華電腦 (CDNS) 2020 Q4 法說會逐字稿

Good afternoon. My name is Rob, and I will be your conference operator today. At this time, I would like to welcome everyone to the Cadence Fourth Quarter 2020 Earnings Conference Call. (Operator Instructions) I would now like to turn the call over to Alan Lindstrom, Senior Group Director of Investor Relations for Cadence. Please go ahead.

Thank you, Rob. I would like to welcome everyone to our fourth quarter 2020 earnings conference call. I am joined today by Lip-Bu Tan, Chief Executive Officer; Anirudh Devgan, President; and John Wall, Senior Vice President and Chief Financial Officer. The webcast of this call is available through our website, cadence.com, and will be archived through March 19, 2021. A copy of today's prepared remarks will also be available on our website at the conclusion of today's call.

Please note that the discussion today will contain forward-looking statements and that actual results may differ materially from those expectations. For information on factors that could cause a difference in our results, please refer to our filings with the Securities and Exchange Commission. These include Cadence's most recent reports on Form 10-K and Form 10-Q including the company's future filings and the cautionary comments regarding forward-looking statements in the earnings press release we issued today.

In addition to financial results prepared in accordance with generally accepted accounting principles, or GAAP, we will also present certain non-GAAP financial measures today. Cadence management believes that in addition to using GAAP results in evaluating our business, it can also be useful to review results using certain non-GAAP financial measures. Investors and potential investors are encouraged to review the reconciliation of non-GAAP financial measures with their most direct comparable GAAP financial results. The reconciliations are available at the Investor Relations section of cadence.com. Copies of today's press release dated February 22, 2021, for the quarter and year ended January 2, 2021, related financial tables and the CFO commentary are also available on our website.

Note that again, today, we are conducting the earnings call from our respective remote locations. (Operator Instructions)

And now I will turn the call over to Lip-Bu.

Good afternoon, everyone, and thank you for joining us today. In the year of unprecedented macro challenges, I'm pleased to report that Cadence achieved outstanding financial results for both the fourth quarter and 2020.

For the year, we achieved 15% revenue growth and 35% on GAAP operating margin with strength across the board and all segments growing by double digit. In a moment, John will present more information for Q4 results and our 2021 outlook. The acceleration of digitization due to the pandemic, coupled with exciting generational trends like 5G, AI, ML, data analytics and hyperscale computing, continue to drive strong semiconductor demand.

Leading hyperscaler, global AI companies continue to aggressively pursue more slow. While 5G, wireless, industrial and automotive verticals are leading the charge on the more than most front. Our intelligent system design strategy triples our aim, and our broad compelling portfolio of chip, package, board and system design solution uniquely positions us to realize these exciting opportunities.

In 2020, we accelerated our momentum and marquee accounts, while expanding our system portfolio through compelling acquisitions, enabling us to win new customers in our targeted vertical segments. Of particular note was the strength of our aerospace and defense business as the digital transformation in this vertical continues. And we expanded and deepened our leadership ship with Northrop Grumman, which includes the use of our custom analog and digital full flow.

Additionally, we are especially excited by the momentum we have with the hyperscalers as they are accelerating their chip and system design activity.

Our engagements with the system companies have steadily risen over the past few years, and revenue from system companies is now closer to 45% of our total revenue. It was a great year for our Cadence Cloud, which now has more than 175 customers. And we further strengthened our partnership with cloud infrastructure and foundry partners.

Now let us review Q4 highlights. Design excellence is a foundational layer of our strategy and includes our core EDA chip design platforms and IP portfolio. Our design -- our digital signoff solutions, offering superior quality of results and faster convergence continued proliferating at market-shaping customers. We engage in well over 100 projects at 5-nanometer and below process technologies, and we are in earlier collaborations under 2-nanometer nodes. Deployment of our full flow accelerated as more than 45 customers adopted our Cadence digital full flow at the most advanced nodes during the year, including MediaTek, Samsung, Micron, Nuvia and a global marquee customer. Based on the superior quality of results, the market-shaping hyperscaler significantly increased the usage of our digital solution. Earlier in the year, a market-shaping automotive semiconductor customer committed to Cadence as its primary EDA vendor for digital design.

Customers are faced with bounding challenges in system verification and software bring-up and is benefiting from our verification full flow solutions that deliver industry-leading verification throughput. Momentum continued on Xcelium, our digital simulator, with several competitive replacements underway as well as expansion and new customer wins. Our hardware family of Palladium or Emulation and Protium for regressions and earlier software development had a strong quarter to finish our best ever year for hardware.

For the year, Palladium Z1 had 24 new customers and 34 expansions, while Protium had 13 new customers and 14 expansion. Demand was particularly strong at AI and hyperscaler customers. Ericsson renewed their commitment to Cadence verification hardware, including both Palladium and Protium. A unique differentiator of the Protium X1 is the common front-end compiler with the Palladium Z1 that enables significant faster bring-up. And about 40% of our hardware business during the year included both Z1 and X1.

Our IP business had a strong year as our focused strategy and a compelling portfolio leverage on the ongoing IP outsourcing trend. We continue expanding the footprint of our leadership DDR and PCIe IP and have several design wins at a 5-nanometer and lower process nodes.

High-speed SerDes is a critical component of hyperscaler infrastructure and several customers had adopted our 112-gig SerDes IP, including [Xsight Labs] who has also demonstrated working silicon. Tensilica has strong loyalties and significant wins as a key true wireless studio and mobile application processor customers. Additionally, the automotive segment had a strong -- have a good momentum which wins in functional safety, radar and digital radio, including an ADAS win, a leading self-driving car company.

In system innovation, we are very excited about our system design and analysis segment, which had a particular strong year with greater than 25% revenue growth. Rising system complexity for advanced 5G, automotive and HPC applications is driving the need for a seamless platform solution across design, simulation and analysis. AWR and Integrand delivered strong great results in 2020, and there is a strong customer interest in our integrated Virtuoso, Innovus, Allegro and AWR microwave office solutions with in-design analysis.

In system analysis, we are executing to our strategy of building out our multi-physics portfolio, offering best-in-class solutions and delivering superior results compared to legacy industry solutions. Strong market adoption continue for our system analysis products. And we are particularly pleased with the growing number of repeat orders with customers, including STMicroelectronics, Realtek and a market-shaping hyperscaler. We triple our system analysis TAM by adding computational fluid dynamic, CFD technology, through the pending NUMECA acquisition, which will bring leading CFD technology and deep domain expertise of Dr. Charles Hirsch and his team for Cadence. NUMECA has over 450 customers, including NASA, Honda and Ford across multiple verticals such as aerospace, automotive, industrial and marine.

For pervasive intelligence, we continue to incorporate machine learning technologies in various tools for improved PPA and faster convergence. Xcelium ML has a successful engagement with several market-shaping customers, delivering up to 5x improvement in regressions throughput. We also progressed on providing IP specifically tuned for AI machine learning applications, especially edge influencing applications.

Cadence continues to invest in fostering innovation and advancing education through endowments at top universities. Building upon previous endowments at Stanford, the University of California at Berkeley and Carnegie Mellon University, we announced a $5 million endowment at MIT Schwarzman College of Computing to promote research in the fields of artificial intelligence, machine learning and data analytics.

Now I will turn it over to John to go over the Q4 results and present our 2021 outlook.

Thanks, Lip-Bu, and good afternoon, everyone.

I'm pleased to report we exceeded all of our key operating metrics for the fourth quarter and fiscal year 2020. We achieved 50% on the Rule of 40 metric for the first time, through a combination of 15% revenue growth and 35% non-GAAP operating margin. Consistent execution against our strategy and double-digit growth across all product categories helped us achieve this landmark, but we also had the benefit of some tailwinds during the fiscal year, which included a 53rd week, a strong second half in China and the recovery of $26 million that we previously thought to be uncollectible.

The pandemic brought many challenges, and I'm very proud of our Cadence team for their compassion and resilience in the face of adversity. Their focus on innovation and customer success resulted in continued acceleration of Cadence's 3-year revenue growth CAGR, which is now into double digits.

Let's go through the key results for the fourth quarter and the year, starting with the P&L. Total revenue was $760 million for the quarter and $2.683 billion for the year. Non-GAAP operating margin was approximately 37% for the quarter and approximately 35% for the year. GAAP EPS was $0.62 for the quarter and $2.11 for the year. And non-GAAP EPS was $0.83 for the quarter and $2.80 for the year.

Next, turning to the balance sheet and cash flow. Our cash balance was $928 million at year-end, while the principal value of debt outstanding was $350 million. Operating cash flow in the fourth quarter was $136 million and $905 million for the full year. DSOs were 44 days, and we repurchased $380 million of Cadence shares during the year.

Before I provide commentary for Q1 and fiscal 2021, I'd like to take a moment to share the assumptions embedded in our outlook. We expect strong revenue growth for the first half, followed by more muted second half growth as we lap some tough second half comps. Our outlook further assumes the pandemic restrictions will gradually ease across the globe this year, resulting in higher T&E expense in 2021 compared to 2020. We've included the expected impact of the pending NUMECA acquisition in our 2021 outlook. And finally, as usual, our outlook assumes that the export limitations that exist today for certain customers will remain in place for all of 2021.

Embedding these assumptions into our outlook for fiscal 2021, we expect revenue in the range of $2.86 billion to $2.92 billion; non-GAAP operating margin of 34.5% to 36%; GAAP EPS in the range of $2.09 to $2.19; and non-GAAP EPS in the range of $2.95 to $3.05; operating cash flow in the range of $900 million to $950 million; and we expect to use approximately 50% of our free cash flow to repurchase Cadence shares in 2021.

For Q1 2021, we expect revenue in the range of $710 million to $730 million, non-GAAP operating margin of approximately 35%, GAAP EPS in the range of $0.55 to $0.59 and non-GAAP EPS in the range of $0.72 to $0.76. And we expect to repurchase $110 million of Cadence shares in Q1.

Our CFO commentary, which is available on our website, includes our outlook for additional items as well as further analysis and GAAP to non-GAAP reconciliations.

In conclusion, Cadence delivered another year of strong operating results, achieving 50% on the Rule of 40 metric for the very first time. We remain focused on driving profitable revenue growth. And at the midpoint of our outlook for fiscal 2021, I'm pleased that we are on track to grow our annual revenue by over $1 billion since 2016, with more than $0.50 of every dollar of revenue growth, dropping through to non-GAAP operating margin over that period.

I would like to close by thanking our customers, partners and our hard-working employees for all that they do, and I'd like to remind them all that their health and safety continue to be our first priority.

And with that, operator, we'll now take questions.

(Operator Instructions) And your first question comes from the line of Gary Mobley from Wells Fargo Securities.

And let me extend that congratulations to a strong start to the fiscal year and a strong execution relative to your Rule of 40, exceeding that by 10 percentage points, and that leads me to my question, John. Could you help us level set what that 35% operating margin in 2020 may have been if you didn't have, I guess, the reversal of the bad debt reserve or bad collectible reserve, the fact that you weren't able to bring on employees at a fast enough rate and then as well the extra week? How does that 35.25% operating margin guide in fiscal year '21 compared to level set fiscal year '20?

Okay. Thanks for the question, Gary. The -- yes, it's a pretty -- it's probably a multipart question. But yes, I guess in terms of the collections, the -- we had a collections windfall of $26 million in the second half, most of which came in Q4. And some of those invoices were very old, which meant that $22 million of the $26 million was recognized in revenue before the end of the year. But -- and that was in relation to, if you recall, in the middle of the year, during the pandemic, we had some smaller customers that weren't able to pay us and we reserved for those. And we helped them out, but our engineers and team at Cadence were determined to help them out, whether they were paying us on us.

But some of those customers turned around and managed to survive and we're assured they would. So that collections windfall kind of distorted maybe some of the revenue timing in the quarters during the year, but didn't really impact -- that $26 million didn't really impact the year because originally, we had it in the year, then we took it out in the middle of the year. And then like that $22 million of that $26 million collected got recognized.

In relation to the extra week, we thought that was going to be $45 million, and it was. But the extra week, of course, was a 2% tailwind for -- in terms of helping us achieve 50% on the Rule of 40 in 2020. And what was the 2% tailwind in 2020 is a 2% headwind now in 2021 when we go from a 53-week year to a 52-week year. And then, I guess, we had a really strong second half in China. I mean when I step back from individual quarters and halves that I would say this, our business is very strong in all regions and across all of the product categories.

The 3-year revenue CAGRs, they're up to 11% now in 2020. And at the midpoint, it's 11% for 2020. At the midpoint of our outlook for '21, that's 11%, too. So we're already off to a very strong start as well in the first half of '21, but it's too early really to say if that's sustainable in the longer term.

So the second half, we have kind of muted revenue guidance. When I look at the second half of 2020 in China. What I noticed was that with a higher-than-normal proportion of upfront revenue in our recurring revenue mix for the region in China, for our outlook in '21, I've assumed we return to our usual recurring revenue mix in the region. And that, along with the fact that we have 1 less week in the second half in 2021 versus 2020, that's what contributes to the like conservative revenue growth outlook for the second half.

Okay. That's very, very helpful, John. I wanted to ask you about the materiality of the software systems analysis product group, including NUMECA. I guess there's perhaps various documentation that NUMECA had roughly $30 million in annual revenue. Is it a pretty good approximation?

And for the overall systems analysis category, how material is this now? Is it somewhere in the ballpark of, what, 1% or 2% of revenue now? And how much is it -- how much is it contributing to your overall growth?

So Gary, yes, the system analysis business is doing great. I mean bookings and revenue grew strongly in 2020. The operating margin profile there is better than EDA, which allows us to invest in the business while we're generating strong incremental margins that improve our overall operating leverage.

In relation to NUMECA, when we issued the original press release, we highlighted that the impact to '21 is pretty immaterial. It's pending. We expect it to close -- is imminent in terms of close, I would expect it to close this week. And as usual, of course, we'd expect purchase accounting rules to initially limit the revenue that we can recognize on NUMECA in the first year. So we expect that to be temporarily dilutive to earnings in 2021, and that's already embedded in our outlook.

So from that perspective, we don't have -- I don't think we have anywhere near as much revenue in '21 as you're expecting.

Your next question comes from the line of Mitch Steves from RBC Capital Markets.

Can you hear me?

Yes, loud and clear.

Okay, yes. So the first one is just on the 45% systems exposure now. I'm just sort of curious about why that's trending up so quickly. About 3 years ago, I think it was closer to like low 40s or 40% which is kind of metric, but now we're suddenly at 45%. And it sounds like that might even go higher. So maybe -- can you help us understand what exactly is going on with that business that's causing it to spike up?

And then secondarily, regarding your 50% flow-through in the operating margin, clearly, that's what you've been doing. But is there any difference between the systems and semis onboarding in terms of profitability for you guys? Those are my 2 questions.

Yes, Mitch, maybe I can answer the first question first about the system company, and then we provide this end-to-end EDA portfolio. And also with our intelligent system design, we are moving up into the system analysis and area and also the packaging side, and then as you can tell, this generation wave in the 5G and also the hyperscale and then autonomous driving. And really, the system company is starting to engage quite actively with us because we're providing a suite of solutions that they're looking for.

So I think it's kind of accumulation of the last 12 years, continue working on it. So we are delighted, we're now getting closer to the 45% of the revenue. And for them, clearly, the system, the packaging, not just the silicon development is critical for them that we can provide an integrated solution for them so that they can design a complex system that they're looking for.

And so we are excited about this opportunity and our strategy investing and behind it. And so we are glad to see the performance.

Yes, Mitch. And to the second part of your question there that -- like you say, that there's a higher than normal profitability, I think, for us in the system analysis business, and it's been growing really fast for us. I mean if I look at incremental margins, I called out in my prepared remarks, this -- where at the midpoint of our outlook for 2021, we're expecting our non-GAAP operating margin to grow by about $550 million over the 5-year period from -- since 2016. And that's about 51% of the revenue growth, if you calculate the revenue growth over that period of time.

And then when I compare the midpoint of outlook for '21 against '19, I did the same exercise, and I got 51% again before the impact of NUMECA. With NUMECA, it kind of takes it just down under $50 million.

Okay. That's perfect. Just to clarify, make sure I understand this correctly, systems analysis. It sounds to me, and maybe I'm reading too much into this, you're seeing like a step function in your 3D Clarity business. Or am I reading too much into this?

What will...

Well...

Yes, more than that I think we also have all these hyperscale guys and then some of the system company service provider. They are quietly also building up semiconductor custom silicon design, and so our entire suite from the design excellence from the EDA, IP and then plus our system analysis, plus upon acquisition we make in AWR and Integrand, we're providing a lot of system for the 5G and also the industrial group.

Your next question comes from the line of Pradeep Ramani from UBS.

Congratulation on a great quarter and a very solid guide. I just had a couple of questions, maybe last one and then I have a follow-up. How are you thinking in terms of the mix you're seeing coming out of China versus the rest of the world?

I mean last quarter, it felt a little bit more towards IP and hardware, but it feels like your growth is now increasingly becoming more broad-based in China and driven by EDA and software as well. Is that a correct read?

Yes. Let me answer that. This is Lip-Bu. So clearly, APAC is a strong growth region for us. And then we've done well in China on Q3 and Q4. And then as you know, China is heavily investing in semiconductor industry. So in some way, we have a broad portfolio of EDA tool, IP and even the system analysis and some of the packaging and then the treated packaging, and it become critical for them. And second, overall, we support their customer globally. And then China especially strong, and we continue to comply with the other export control requirements. But meanwhile, so far, knock on wood, we see strong momentum over there.

Pradeep, I would add there that -- like if you step back from looking at any individual quarter or half and you look at the 3-year CAGR, I think our growth is accelerating across all regions in our outlook for '21.

Great. And for my follow-up, you had a very strong year on IP. How do we think about the sustainability of growth in IP going forward, given, of course, there's longer-term drivers, but just coming off of such a strong year?

Yes. So I think it's a good question. IP tend to be lumpy. But so far, we like this IP outsourcing trend. And we have a strong portfolio in terms of DDR, PCIe and also the SerDes is a must-have for the hyperscale and also some of the Tensilica in terms of audio and also the automotive sector.

So overall, we are delighted last year is a strong year. But as you know, it's a very lumpy upfront. And so we will continue to focus on that.

Your next question comes from the line of Jason Celino from KeyBanc Capital Markets.

The first question, NUMECA acquisition, I know it's pending, but maybe can you speak to the -- what is so attractive about that business. I know CFD, the pillar physics part of the market. And then make versus buy there, I mean what factors went into acquiring versus just building in?

Anirudh, do you want to take this one?

Yes, definitely. Yes, that's a good question. So as you know, we are pretty excited about our move into system analysis, which is driven by our expertise in computation software or numerical mathematical software. And we have a lot of expertise in that organically as we saw in the development of Clarity, which is the 3D EM Solver that has been generating a lot of good results.

And the second thing we are excited about is the customer synergy. A lot of customers -- as we discussed, we have a lot more system companies and they are asking for the system analysis capability, okay. And then the question becomes whether to organically develop or to acquire. And we look at the space. And in terms of system analysis, CFD is a very critical area. It's one of the largest market segments with a lot of vertical application. And NUMECA has very good technology with more than 450 customers.

So we are pleased to welcome them to Cadence with this pending acquisition. And I think that can be used as a basis to expand further, expand more in R&D with our competition software strength and expand more with our customer base. So we are very happy to use them with their expertise in CFD to expand into this very, very exciting area for us.

Okay. And then my quick follow-up with this is I think you mentioned that it was a 3x TAM expansion with the acquisition in the CFD. Does that 3x TAM expansion also include other physics like structurals? Or is that just CFD alone?

So in terms of system analysis, that's a pretty big segment, and it's about $6 billion, we estimate. And the other good thing about it is that it is growing rapidly. There is always more need for more and more simulation. What we estimate is clarity and sales deals is about $800 million in addressable market, which is EM and thermal and CFD is about 1.6. So CFD triples our TAM from $800 million to about $2.4 billion.

Your next question comes from the line of Jason Ader from JPMorgan.

Great. The first question is on the 175 cloud customers. What does that mix in terms of systems versus semiconductor companies look like?

Yes. We don't have a breakdown on that. But clearly, we are very excited in our customer. It's -- clearly, our solution provides the flexibility and also different use model and either it's Cadence-managed. And so overall, we're providing EDA software and Palladium platform.

The most important is providing the productivity and scalability. So I think overall, we are excited of 175 customers. We didn't break down in terms of semiconductor or system company. But overall, both are strong for us.

Okay. And then John, you mentioned that not expecting China to be as active in upfront revenue in 2021. So any particular reason? I mean was it kind of politically driven why that particular geography had a bigger mix of hardware and IP in 2020?

No. Well, it's when we have more hardware in any one quarter, in any one region, you're going to have a higher mix of upfront revenue in that region in that quarter. And we saw that with a strong Q3 for -- in the functional verification space and a lot of that strength within China. And when I was looking for signs of a pull-in in Q3 and -- to try and figure out that -- was there some acceleration of purchasing that came out in '21 or did it come out of Q4 where people just preparing with the pandemic, where they're just trying to get their hardware delivered earlier for fears of delays due to the pandemic. But -- and then what we looked at, I was waiting for a view of the pipeline. Pipeline for '21 is very strong.

And when I look at that and compared it to what had happened, what was different about the second half of 2020 for China, what I did notice is because of that hardware that we had a higher mix of upfront revenue in the region in the second half than we would normally see.

I'm not sure if that's sustainable. I mean it looks like we're off to a strong start again in Q1. But for the purposes of determining an outlook for 2021, I assume -- I thought it was safest to assume a return to our usual recurring revenue mix, which could prove to be conservative for the second half. But I'd rather go out with outlook assuming that we return to the usual recurring revenue mix than assume that that growth that we saw in the second half of 2020 continues.

I mean if it does -- when we have increased visibility into revenue in the second half, we can uplift -- we can update our outlook at that time when we see it. But I just -- I didn't want to go with too strong a second half right now until I have a clear visibility into the pipeline in the second half.

Next question comes from the line of Joe Vruwink from Baird.

Great. John, you've been referring to the tables in the release. And I'm interested in the one that breaks it down, 3-year CAGR adjusted for the extra week by product group. If you think into 2021 -- and again just adjusting for the weak comparison -- would you expect kind of similar dynamics between the product segments where IP and system design should remain kind of the fastest growers? Or are there any new developments at a product level to consider as 2021 goes on that might influence some of the trends you've seen?

Good question. I mean we're not guiding by individual product category. But if you see -- if you look at the 3-year CAGR for '18, '19 and '20 on that table, you see that don't change dramatically year-over-year, particularly on the 3-year CAGR view. But the ones you call out, IP and system design analysis are kind of the smaller dollar values, but -- so they have the benefit of growing from smaller numbers. And in the past, we've seen those grow faster.

But in relation to new products, I think that the amount of innovation that we've seen in the new product releases and preparation that we've seen from the R&D group, and they haven't slowed down at all. If anything, they're accelerating in their new product development even through the pandemic, which has been fantastic. But typically on an earnings release, we wouldn't announce any new products, you'll see those at CDNLive! Would probably start around late March, April time.

Okay. Okay. Great. And then this next question might be product development related. But if I heard Anirudh correctly, systems design is maybe a $6 billion total opportunity. And given the solvers you've introduced or the new CFD being acquired, you're up to $2.4 billion. And so are there other categories or additional things that you have in mind that we can maybe expect you growing into over the next few years?

Yes, that's a good point. And first of all, I want to highlight that we are building out this multi-physics platform. And CFD is a pretty big segment of that market. So we started with finite element and electromagnetics and then now go to CFD, and there's a lot of commonalities there. So I believe that itself presents a lot of opportunity to us, like we mentioned, tripling our addressable market. There might be more in the future. We always look for that. But at the same time, the current collector magnetics and CFD already has a large market segment that we're busy addressing.

So the answer is yes, if opportunities come, we will systematically expand, but already, that's a good opportunity for us.

Your next question comes from the line of John Pitzer from Crédit Suisse.

Congrats on the solid results. John, as you rightly pointed out in your preamble, if you adjust for the extra week and some of the unexpected revenue in '20, your initial guide for '21 is already embedding sort of double-digit growth. And I know you said that it's too early to say if that's the new norm. But I guess I'm kind of curious, what are you looking at to be able to make that call that this might be the new normalized growth rate?

And Lip-Bu, as you talk about potentially a double-digit secular growth rate, what do you think is driving that? Is that sort of specific to Cadence and your strategy? Do you think this is core EDA customers? Do you think it's the proliferation of customers into nontraditional areas like hyperscale, systems and autos? If you can give us a little bit of help on that, it would be appreciated.

So John, I can take your -- the first part of your question, and then I'll pass it over to Lip-Bu. In relation to the strong first half guide, strong first quarter guide, most of our revenue is software, and most of it's recurring revenue. I mean we expect recurring revenue to make up 85% to 90% of our revenue for the year. So it's very, very predictable and very, very predictable, certainly in the near term.

Where I had some difficulty was projecting what upfront revenue might be out in Q3 and Q4. So the outlook for the second half is more conservative than the -- the first half is very, very predictable in terms of great visibility into the pipeline. We're off to a really strong start in Q1 already on the upfront business. And so much of that revenue is recurring in the first half that -- yes, we're seeing accelerating revenue growth.

The challenge in the second half, of course, we're lapping very tough comps against the second half of 2020. And I just don't have the visibility into the upfront pipeline for Q3 and Q4 yet.

Yes. On the second question, let me describe. I think it's kind of exciting about this renaissance of semiconductor, with the 5 generation waves happening at the same time, AI, machine learning, all about data, data analytics, 5G and then the cloud infrastructure had to change because all this massive data coming in and then how do you address that network scaling and then storage is aggregation.

So it's a lot of changes to infrastructure and there's a lot of fueling, a lot of design activity. So that's why we see a lot of design activity, not just from the semiconductor players. Now the system player and then the whole Industry 4.0, they all move into AI, machine learning, the digital transformation that drive a lot of semiconductor. So overall, the industry is moving in the right direction to us.

And then meanwhile, I think, clearly, Cadence's Intelligent System Design strategy really tie in really well from the design excellence, providing the tool in IP. Now we move to the next level of system innovation that Anirudh talked about the system analysis and the CFD and the physics or multi-physics model. Then the next thing is move up into the Pervasive intelligence that's using AI machine learning, data analytics, drive all the vertical industry, that is major transformation going on. So I think we are well positioned to capture that opportunity.

And then guys, my second question is, clearly, when you look at China as a region, from Q4 of '19 to the back half of last year, it almost doubled as a percent of revenue. And John, you clearly talked about potentially some onetimes that came in, in the back half of the year that you're not embedding sort of in the guide for '21. I'm just kind of curious, as we think about currently China at about 17%, what's kind of core embedded as a percent of revenue in your '21 guide, realizing that that's subject to change.

And if you think longer term with China's aspiration, is this about the right level as a percent of revenue or would you expect this on a secular basis to continue to grow over time?

Well, again, individual quarters, I wouldn't focus too heavily on any one individual quarter or even half. I mean it was a tremendous second half for us in China. And 2020 was a strange year, I mean with the pandemic, with the extra week in Q4. I mean you'll see in Q3, we had a lot of strength in our functional verification group. In Q4, you'll see the software businesses performed really, really well. Of course, it had the benefit of the extra week with a 14th week in your normal 13-week quarter. But -- and we have 1 less week in the second half of 2021.

We're very, very pleased with the growth we're seeing in China. It looks like it's continuing on. We're seeing that strength continue on into Q1. The pipeline is strong. We're seeing that in the first half. I just wanted to be conservative in the outlook for the second half until I have better visibility into second half upfront revenue pipeline.

Your next question comes from the line of Tom Diffely from D.A. Davidson.

Maybe first, just a clarification for John. When you look at your expectations for 8% revenue growth and 8% OpEx growth and the lack of leverage there, did you say that was primarily due to just the NUMECA acquisition, the increased cost?

No, it's not. Certainly not primarily due to the NUMECA acquisition. I think in terms of expense growth, our headcount is up 8%. Our head count increased by 8% during 2020. And we're investing heavily again in hiring in 2021.

NUMECA will come into the mix, we hope, for the second half of Q1 and then the remainder of the year from an expense perspective. We expect merit to hit this in -- merit increases will be July again. So that will be -- you'll see a slight uptick there in expense in Q3 and Q4. And the growth in expenses through the year that would typically come from that is being offset by some more efficient infrastructure spend that we have planned as the year progresses.

If you look at our 10-K that we filed, you'll see that we initiated a restructuring plan in Q4 2020 to optimize our spend on infrastructure. But -- and the expense mix in 2021, I think if you compare '21 versus '20, you'll probably see in '21, we're spending more on people and less on places than we were in '20. That's kind of driving some of the margin profile.

Okay. That's very helpful. And then a follow-up for Lip-Bu, when you look at the cloud, you talked about 175 customers, what do you think the long-term adoption is for EDA with the cloud? Do you think some of your really large customers will move the majority to the cloud at some point? And if they do, what does that do to your cost structure?

Yes, very good question. This is still at a very early stage. We are very encouraged with the 175 customers. And clearly, we want to make sure that our initial product in terms of system analysis, we can be cloud-native because we start from scratch. And then some of the EDA tool, we have different stage of moving to cloud-native. And then, really, the holy grail is basically drive the productivity efficiency for our customers.

And then our partnership with the hyperscale partners and infrastructure partners and also the foundry partners are critical in this effort. And so right now, we have across small, medium, large customer embracing on cloud. And truly, I think, if you can imagine we have the unlimited usage of machine server at your exposure and then so that you can really drive the performance, the productivity significantly. So I think that is a trend.

We are just in the beginning early stage, but we are very encouraged with the customer support. In order to continue to grow, clearly, our partnership with the hyperscale partners and also the foundry partners are critical because of the PDK or the different solution they have. We want to make sure that we work with them together and support our customer and also addressing the need of customers.

Your next question comes from the line of Gal Munda from Berenberg Capital.

The first one, I just wanted to follow up briefly on NUMECA and more generally on your ambitions within the kind of mechanical simulation space. If you think about the convergence that's happening with NUMECA, now you said you have the CFD, you have the finite element analysis as well. So like how far do you think your portfolio is, based on what you have to be able to take it where you want it to be, say, in R&D terms versus potentially doing more of those tuck-ins in the CAE space?

Yes, that's a good question. Let me take that. This is Anirudh. So like I mentioned, this move is driven by 3 factors, right, our strength in competition software; our customers asking for more and more system analysis, system design capabilities, like Lip-Bu mentioned; and the overall need for simulation.

And I think we are still in the early innings, but we feel confident about this space, as demonstrated by our results in Clarity and other products. And CFD is a pretty important area, I mean I think one of the biggest segments in system analysis. The good thing about CFD is it has a lot of vertical applications, all the way from automotive, aero and defense, all the way to medical. So I think that's a pretty significant expansion of our platform. And we are happy with this progress. We are patient. And like John mentioned, this segment is profitable. So we will continue building across this and keep providing the best solution to our customers.

Makes sense. And then just as a follow-up, maybe just expanding a little bit on the growth drivers of the businesses, Lip-Bu, you mentioned, there's Moore's Law-based growth at the leading edge customers, and there is more than Moore's, which is kind of the system side.

Considering the investments you're making on the system side, can you just comment a little bit on the growth, how it -- maybe 2020, how it corresponded between the leading edge and the systems companies in terms of that contribution to get you to, whatever we want to say, 13% or 15%, depending on which number we're looking at?

Yes, it's a very good question. And so I think we are very well positioned to be able to do both. And then clearly, the hyperscale, mobile, they are driving a lot into the Moore's Law. And as I mentioned that we are very delighted. We have more than 100 projects on 5-nanometer and below process node development with our customers. And we have earlier collaborations on 2-nanometer process.

So I think this is a very advanced and we're delighted, and we are very honored to work with the marquee customers. They are pushing the envelope, and we are really excited about that development.

On the other hand, we are so well positioned in our custom mixed-signal, analog and packaging. Like the 5G, the wireless industrial group, they are really moving into this mixed-signal and, clearly, the most and more. And so that we are very well positioned, and we doubled down with some of this AWR and then the Integrand acquisition driving some of the RF microwave requirement, especially in the antenna side of some of these 5G integrations. And then we have really good solution for that, and customers see the benefit of working with us.

So I think we are very well positioned on both side. That's why we highlight that from our portfolio point of view, and both engines are really taking off, and we are very excited, engaging with all the best company to work with us.

Your next question comes from the line of Rich Valera from Needham.

Let me add my congratulations on the Rule of 50 last year. Question, I think it's for Anirudh on the system analysis go-to-market. Just wondering, today that you have been doing that through the standard sales channel, and what do you -- do you have any overlay sales and/or specialized applications for system? And how are you thinking about scaling up any kind of dedicated system resources as you scale up that business?

Yes. Thanks, Rich. That's a great question. And of course, one is developing great products and the other is go-to-market or helping our customers deploy those products. And the second part is as important as the first part.

And especially for the system business, there are multiple facets to it. In the semiconductor business or the big large system companies, we operate mostly through a direct channel, okay. But if you remember, we also have our Allegro business, which is our packaging business. We do quite well in PCB and packaging with Allegro. And Allegro does have a -- already, we do have an indirect channel. And like Lip-Bu mentioned earlier, cloud is important going forward.

So for the system business, I envision 3 channels. We will -- we are building 3 channels. One is the direct channel because a lot of the big companies and big system in semi companies, anyway, want system tools, like you mentioned. And expanding our indirect channel like we have for Allegro and expanding it to overall system design analysis, and then a SaaS channel, which -- or cloud-based channel for -- especially for smaller companies that don't have a lot of data centers and are more attractive to use on the cloud.

So as we go to the system design and analysis, we do have to build out these 3 facets of our go-to-market. And we are working all of them in different stages. But the strength of our direct channel helps us with our big customers. And then indirect in SaaS can help us with some of the smaller long-tail customers. But that's a very important part of the overall go-to-market strategy.

Understood. And then, John, a follow-up for you. You'd given a backlog reserve number last quarter of $58 million, which I think has come down from $70 million originally. And I'm assuming that came down materially in the fourth quarter. Can you tell us where that is today?

Yes, of course, Rich, that we collected -- we had a collections windfall of $26 million in the second half out of that $70 million, that I think the majority of the balance of the $70 million is lost. I mean those customers have -- many of those customers have closed their doors. I think we're in single digits in millions in terms of the ones we're left tracking that could potentially be recoverable. But right now, in my outlook, I'm assuming it's not recoverable. But you're talking maybe $9 million, $10 million left out of that $70 million, that could possibly be recovered. But like I say, right now, I'm assuming it doesn't get recovered. We recovered $26 million of the other $60 million.

Your final question comes from the line of Jay Vleeschhouwer from Griffin Securities.

I'd like to direct both of my questions to Anirudh. First, we've heard for years, of course, from the EDA companies of ongoing design activity among your customers that that's remained robust. We've also heard of the general direction of design style or types from general purpose chips more to specialty kinds of designs. What we don't often hear about is how customers' design methodology may be evolving.

And I'm wondering if there are any issues that you might want to talk about regarding, for example, the hierarchical versus flat, which has been a consideration for chip design now for 2 decades or more. Is -- are there any new implications or issues for that as it might pertain to or affect any of your EDA tools or new technologies, such as iSpatial or design space exploration?

And then relatedly, there were a number of questions this evening already regarding system analysis and your broader ambitions in computational software. One thing we see on the other side of engineering software, the world you're approaching towards, are unified data platforms and common platforms for data management and the like, which you don't really seem to have. You do have multiple stacks for your various functional areas. Could you talk about how you're thinking about unifying or moving to a more common platform of that kind?

Yes, Jay, these are a great question. Well, first of all, we want to make sure that we have best-in-class products, right, so whether that's digital design or verification and system analysis.

So we always, as you know, this is technology business, best product wins. And at the same time, have a unified platform to go with it. And this is how we did digital implementation. This is how we did verification. So both parts are important to have best-in-class products, whether that's in finite element, electromagnetics or CFD. And then there are opportunities to combine them in a much more unified way, especially around data, like you mentioned.

So I completely agree that the need for unification, but we also, at the same time, I want to make sure first that the products are best-in-class. We want a unified team, but each player has to be a strong player. But there are a lot of opportunities for data platforms and analytics like Lip-Bu also mentioned earlier.

Now in terms of your first question of new methodologies, I mean there are several things happening. If I have to pick 1 or 2 key things that are happening, what we see with the customers, so one big trend, which you probably already know is this move to 3D-IC. I think this is going to be significant for the next several years. And it just changed a lot of methodologies in our customer base. And this is multiple high-end digital chips or even analog RF chips on a package. And that, I think, is driving a lot of change, especially in the cloud, hyperscaler companies all the way to consumer company.

And we do have, historically, a lot of strength in package -- packaging with Allegro. Of course, we have a leadership platform, an advanced node with Innovus and then now the analysis tools with Clarity and Celsius because thermal is a big thing for 3D-IC. So I think this whole methodology of 3D-IC integration is one significant change that we do see in the customer base. And we are very excited about that because I think Cadence is in a unique position because of historical strength in analog and packaging and digital and the new strength in system analysis. And hierarchically, you mentioned, I think that's always a big thing as things get bigger.

And if you look at 3D-IC, this is another way to do hierarchical because if you have 4 chips on a package, you don't have to redesign all 4 of that. This is reuse at a much higher level. So -- but I do think 3D-IC is a significant trend that is going to happen, and we are well positioned for it.

I will now turn the call over to Mr. Lip-Bu Tan for some closing remarks.

Thank you all for joining us this afternoon. I'm very excited about the growing market opportunities and business momentum going into 2021.

Our Intelligent System Design strategy is playing out very nicely as we benefit from new opportunities in design excellence, system innovation and pervasive intelligence and an expanded auto addressable market.

We excelled in the challenging year, thanks to the deep partnership with our customers and our partners and the strong commitment of the outstanding Cadence team.

And lastly, on behalf of all our employees and our Board of Directors, we give heartfelt thanks to those on the front lines who continue to work tirelessly in their effort to put this pandemic behind us.

Thank you all for joining us this afternoon.

Thank you for participating in today's Cadence Third Quarter 2020 Earnings Conference Call. This concludes today's call. You may now disconnect.