Earnings Call Transcript - CDNS Q2 2023

Operator, Operator

Good afternoon. My name is Lisa, and I will be your conference operator today. At this time, I would like to welcome everyone to the Cadence Second Quarter 2023 Earnings Conference Call. All lines have been placed on mute to prevent any background noise. After the speakers’ remarks, there will be a question-and-answer session. Thank you. I will now turn the call over to Richard Gu, Vice President of Investor Relations for Cadence. Please go ahead.

Richard Gu, Vice President of Investor Relations

Thank you, operator. I would like to welcome everyone to our second quarter of 2023 earnings conference call. I'm joined today by Anirudh Devgan, President and Chief Executive Officer; and John Wall, Senior Vice President and Chief Financial Officer. A webcast of this call and a copy of today's prepared remarks will be available on our website at cadence.com. Today's discussion will contain forward-looking statements, including our 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. For information on factors that could cause actual results to differ, please refer to our SEC filings, including our most recent forms 10-K and 10-Q, CFO commentary and today's earnings release. All forward-looking statements during this call are based on estimates and information available to us as of today, and we disclaim any obligation to update them. In addition, we will present certain non-GAAP measures, which should not be considered in isolation from, or as a substitute for, GAAP results. Reconciliations of GAAP to non-GAAP measures are included in today's earnings release. For the Q&A session today, we'd ask that you observe a limit of one question and one follow-up. Now, I will turn the call over to Anirudh.

Anirudh Devgan, President and CEO

Thank you, Richard. Good afternoon, everyone, and thank you for joining us today. Cadence delivered excellent financial results for the second quarter of 2023 with strong ongoing customer demand for our innovative technology. We exceeded our guidance on all key metrics and are raising our financial outlook for the year yet again, resulting in 14% year-over-year revenue growth and 19% non-GAAP EPS growth. John will provide more details shortly on both our Q2 results and the updated outlook for the year. With its unparalleled promise, Generative AI is beginning to make a significant impact globally. Our dedicated focus on AI over the past several years combined with our computational software expertise and invaluable data that lies at the core of AI uniquely positions us to deliver to this tremendous potential of this transformational knowledge. With escalating design complexity, increasing design starts, and a growing talent shortage, AI-driven design automation is crucial in empowering customers to fully realize their innovation potential. Our customers are increasingly adopting our chip, package, board, system, Generative AI portfolio as they're achieving exceptional quality of results and productivity benefits with these solutions. Customers are ramping up their R&D spend in AI-driven automation, opening up significant opportunities for Cadence. Our solutions are enabling marquee AI infrastructure platform companies to deliver their next generation compute, networking, and memory products. For instance, in his Computex keynote earlier this quarter, Jensen Huang of Nvidia noted that Nvidia is a big Cadence customer and commented on the expanding strategic partnership between Nvidia and Cadence to accelerate EDA, system analysis, AI, and digital biology. Cadence has successfully collaborated with Tesla on the development of their game-changing DOJO AI supercomputer. Tesla utilized a broad array of Cadence solutions across digital, custom analog, verification, 2.5D and 3D IC and system analysis for developing DOJO chips and solutions. We are very excited to extend our partnership to Tesla's next generation DOJO and FSD platforms. Our customers like Tesla are able to leverage the power of revolutionary Cadence Cerebrus Generative AI technology to optimize the quality of results of their groundbreaking AI chips and related solutions. And from a product perspective, we began by providing AI-driven solutions for core ED applications, then expanded our AI portfolio to include system design and analysis and plan to extend it to life sciences in the future. With AI as their underpinning, other generational trends such as hyperscale computing, 5G, and autonomous driving continue to spur robust design activity across semi and system companies, creating ample market opportunities for our differentiated technology portfolio. Now let's talk about our key highlights for Q2. In Q2, we deepened our long-standing partnership with a marquee electronic system company through a broad-ranging expansion of our core EDA, hardware, IP, and systems portfolio. Ambulation and prototyping have become a must-have part of the chip tape-out and software bring-up flows and secular demand for our hardware platforms drove our verification business in Q2 to a 27% year-over-year revenue growth. Following a record 2022 and Q1 '23, our Palladium Z2 and Protium X2 hardware platforms delivered a record Q2 as market demand accelerated for these best-in-class solutions. With 14 new customers and 45 repeat customers, more than half the orders during the quarter included both platforms. Demand for our hardware solutions was broad-based with particular strength seen in AI, hyperscale computing, and automotive segments. A global communication services leader successfully deployed the Z2 and X2 systems to significantly accelerate the development of their data center chips for internal use, including those designed for AI applications. Our hardware platform enabled them to accelerate their verification workload and software bring-up, enabling first-pass silicon and software success. Verisium, our AI-driven verification platform that's built upon the ZI database and natively integrated with our verification engines is gaining traction at market-shaping customers. For instance, a large mobile chip company is working with us on developing a new Verisium regression optimization app and a Japanese semi customer is actively deploying Verisium apps across his automotive SoC and already saw a four times improvement in the identification of erroneous source code check-ins. Our digital IC business had another solid quarter with 15% year-over-year revenue growth largely driven by the proliferation of our digital full flow, especially at the most advanced node at market-shaping customers. Our Cadence Cerebrus solution leverages breakthrough AI-driven technology to explore the entire design space and automatically optimize the digital full flow to deliver transformational results. Adoption and proliferation of Cadence Cerebrus accelerated. And with multiple new marquee wins, it is now deployed at eight of our top 10 customers. A leading hyperscaler doubled their run rate through a major expansion that included our digital full flow and Cadence Cerebrus. Another market-shaping hyperscaler successfully used our digital full flow and Cadence Cerebrus to achieve a 16% leakage power reduction on their latest custom silicon. Through analysis, efficiency, and significantly reduced iterations between RTL and implementation. On IP, our scalable and profitable growth strategy continues benefiting from ongoing outsourcing trend, as well as the opportunities offered by expanding foundry ecosystem. Demand for our design IP was strong, led by our multiyear agreement with Samsung Foundry to expand the availability of our design IP portfolio for Samsung's advanced process technology. We signed a definitive agreement with Rambus to acquire their five IT asset along with the talented team. The addition of their leading HBM GDDR5 solutions and SerDes IP enhances our established IP portfolio, providing complete subsystem solutions for demanding networking, hyperscaler, and AI application. Our system design and analysis business continued its strong momentum in Q2, delivering 23% year-over-year revenue growth. With Moore's Law slowing down and chip complexity and cost increasing, companies are looking to multi-die, 3D IC, and chiplet-based architectures to achieve better performance and greater cost savings. In Q2, we expanded our collaboration with Samsung Foundry through the delivery of reference flows and package design kits based on our uniquely differentiated integrity 3D IC platform. The industry is the only unified platform that includes system planning, packaging, and system-level analysis in a single cockpit. Our multi-physics solutions, boosted by our optimality Generative AI technology, leverage differentiated system simulation and optimization techniques to deliver superior results. Our CFD portfolio, which includes recently acquired high-fidelity simulation technology, won large renewal and add-on business with a top Aerospace and Defense company. We were pleased with the new win and growing repeat orders for our multi-physics portfolio from customers across multiple end markets. In summary, I'm pleased with our Q2 results and the continuing momentum of our business. AI-driven automation in chip and system design offers massive opportunities for Cadence over the long term that seamlessly align with our computational software, core competence, and intelligent system design strategy. We continue to invest in our world-class EDA system design and analysis, and AI capabilities to deliver cutting-edge innovation to our customers and partners. Now I will turn it over to John to provide more details on the Q2 results and our updated 2023 outlook.

John Wall, Senior Vice President and CFO

Thanks, Anirudh, and good afternoon, everyone. I'm pleased to report that Cadence achieved strong results for the second quarter of 2023, driven by the broad-based strength of our business. Our customers are increasingly adopting our Generative AI portfolio as they are achieving exceptional quality results and productivity benefits with these solutions. Here are some of the financial highlights from the second quarter, starting with the P&L. Total revenue was $977 million. GAAP operating margin was 30.7%, and non-GAAP operating margin was 41.8%. GAAP EPS was $0.81, and non-GAAP EPS was $1.22. Next, turning to the balance sheet and cash flow. The cash balance at quarter end was $874 million, while the principal value of debt outstanding was $650 million. Operating cash flow was $414 million, and we used $325 million to repurchase Cadence shares in Q2. We are increasing our outlook for the remainder of the year due to continued broad-based strength across our technology portfolio. Demand for our functional verification hardware solutions remains particularly strong, and as a result, our updated outlook for the second half of the year reflects approximately 15% revenue growth compared to the second half of last year. Before I provide our updated outlook for the year and our expectations for Q3, I'd like to highlight that our outlook contains our usual assumption that the export control regulations that exist today remain substantially similar for the remainder of the year. Our updated outlook for fiscal 2023 is revenue in the range of $4.05 billion to $4.09 billion. GAAP operating margin in the range of 30.2% to 31.2%. Non-GAAP operating margin in the range of 41.2% to 42.2%. GAAP EPS in the range of $3.35 to $3.41, and non-GAAP EPS in the range of $5.11 to $5.5. Operating cash flow in the range of $1.3 billion to $1.4 billion, and we expect to use at least 50% of our annual free cash flow to repurchase Cadence shares. A large number of hardware systems are slated for delivery in late September and early October. For the purposes of providing our outlook for Q3, which ends on September 30th, we thought it was prudent to assume that the vast majority of those hardware deliveries fall into October and Q4. With that in mind, for Q3, we expect revenue in the range of $990 million to $1.01 billion. GAAP operating margin of approximately 29%, non-GAAP operating margin of approximately 40%. GAAP EPS in the range of $0.76 to $0.80, non-GAAP EPS in the range of $1.18 to $1.22. And we expect to use approximately $125 million of cash to repurchase Cadence shares. As usual, we published a CFO commentary document on our Investor Relations website, which includes our outlook for additional items, as well as further analysis in GAAP to non-GAAP reconciliations. In conclusion, we delivered a strong Q2 and first half of the year. With the increase in our outlook, at the midpoint, we now expect revenue growth for the year to exceed 14%, which would take our three-year revenue CAGR to approximately 15%. And non-GAAP EPS growth of approximately 19%, which would result in an average annual growth rate of 22% over the past three years. As always, I'd like to close by thanking our customers, partners, and our employees for their continued support. And with that, operator, we will now take questions.

Operator, Operator

Thank you. Our first question comes from Jason Celino with KeyBanc Capital Markets.

Jason Celino, Analyst

Great. Thanks for taking my question. John, Anirudh. Good quarter, I think if we take a step back, you know, up 14% for the year, still pretty impressive. Sounds like hardware had another record quarter and was up 27%, but I think some investors will still wonder, you know, could it have been even better? Maybe could you speak to any areas that maybe weren't as strong that you were hoping, or how do we kind of level set the performance?

Anirudh Devgan, President and CEO

Yes. Hi, Jason. Good question. In general, we manage our business for the long-term, and we are pretty positive with the results. For the full year, like you mentioned, it's more than 14% revenue growth and earnings are growing at 19%. And then our strategy of doing EDA plus STA plus AI with the computational software expertise. So I think overall, we're pretty pleased with how all the businesses are performing. And, you know, we continue to drive on all fronts, working with customers and partners to deliver the right solutions.

John Wall, Senior Vice President and CFO

We are very satisfied with our performance in the first half of the year. One unusual aspect of the numbers is that operating cash is flat compared to our guidance from last quarter. This is because we are focused on long-term management of the business. We have had opportunities to build inventory due to strong hardware demand and have also received larger discounts for making upfront payments. We have included this in our guidance, which will benefit our gross margins in the coming years, but it does not assist with operating cash for this year.

Jason Celino, Analyst

Got it. And then my quick follow-up, the IP acquisition from Rambus. Is it included in guidance? And then maybe can you speak to strategic rationale? I guess what goes into the decision, build versus buy as related to IP? Thanks.

John Wall, Senior Vice President and CFO

Yes, Jason. So everything we know is included in our guidance, but it's immaterial to us for the year.

Anirudh Devgan, President and CEO

Yes. And it's a great opportunity for us to acquire a very good set of IPs and a talented team. And, you know, what is particularly good in this case is that it is an existing commercial operation. You know, sometimes we have an opportunity to acquire teams from, you know, other semiconductor companies, but in this case, it was an existing commercial operation with key IPs like, you know, HBM and GDDR. So we took that opportunity, and we're glad to, you know, partner with Rambus here, and it really helps, you know, IP for AI and 3D IC with HBM and GDDR IPs. So overall it’s a good acquisition, and we continue to invest in all our businesses, you know, including IP.

Jason Celino, Analyst

Excellent. Excellent. Thank you both.

Operator, Operator

We'll take our next question from Charles Shi with Needham & Company.

Charles Shi, Analyst

Hi, good afternoon. Thank you for taking my questions. First, I want to ask about bookings, and then I have a follow-up on hardware. John or Anirudh, are you both still seeing strong bookings potential in the second half? I noticed that your implied bookings are already up in Q2, but I think people are used to seeing bookings over $1 billion per quarter. When do you think we can return to that level? Do you still expect strong bookings in the second half, similar to the first half of 2022? Thank you.

John Wall, Senior Vice President and CFO

Yes, Charles, great question. Yes, we'd expect a stronger second half for bookings. As we said last quarter, we had very few contracts expiring in the first half of this year, and we have more contracts expiring in the second half of 2023. So we'd expect bookings to be higher in the second half compared to the first. When we look at backlog, we filed our 10-Q, by the way. So you'll see our backlog at the end of Q2 dropped from $5.4 billion to $5.3 billion, and RPO at the end of Q2 dropped from $5 billion at the end of Q1 to $4.9 billion. But the current value of that RPO, the annual value, as you know, backlog and RPO is a combination of the annual value multiplied by the time essentially that you've contracted with customers. And our current RPO is up from last quarter; it was $2.7 billion last quarter, and it's up to $2.8 billion this quarter. So we've improved the annual value, and that's typically what we focus on, making sure we're improving the annual value. But yes, we're very, very pleased with the bookings performance for Q2. It was slightly stronger than we had anticipated or expected. And in the second half, we think it will be better again.

Charles Shi, Analyst

Yes. Is it accurate to say that Q1 represented the lowest point in terms of bookings, and we have moved beyond that?

John Wall, Senior Vice President and CFO

Yes, yes, Q1 was definitely a low bookings quarter for us. Q2 was a lot stronger.

Charles Shi, Analyst

Got it. I have a follow-up question about hardware. You mentioned that for Q4, you expect a significant amount of hardware revenue. However, when looking at your full-year guidance, Q4 doesn't show a considerable increase, which implies the guidance for Q4. Given your statement about expecting strong bookings, I would assume that the recurring part of the business should see substantial sequential growth in Q4. Additionally, considering the hardware you anticipate for Q4, could you clarify the discrepancy between your Q4 guidance and your comments on hardware and bookings? Thank you.

John Wall, Senior Vice President and CFO

Yes, Charles, I think you're referring to the step-up from Q3 to Q4 in our guide. So we're guiding about $1 billion of revenue in Q3 with a balance of just over $1.70 billion in Q4. Now that's slightly exaggerated the Q3 to Q4 step-up. When we originally looked at the top-down forecast for the second half, there was a lower step-up in revenue mainly due to software increases in annual value and in software bookings that we expect. But when we looked at the hardware and the timing of when hardware deliveries happen and there's so many that fall late in the quarter, we thought it was prudent to assume that those bookings, we should include them in our implied Q4 guidance, so to speak, rather than including it in Q3 because I’m not really sure about the timing of those. Yes, Charles, I think you're referring to the step-up from Q3 to Q4 in our guide. So we're guiding about $1 billion of revenue in Q3 with a balance of just over $1.70 billion in Q4. Now that's slightly exaggerated the Q3 to Q4 step-up. When we originally looked at the top-down forecast for the second half, there was a lower step-up in revenue mainly due to software increases in annual value and in software bookings that we expect. But when we looked at the hardware and the timing of when hardware deliveries happen, and there's so many that fall late in the quarter, we thought it was prudent to assume that those bookings, we should include them in our implied Q4 guidance, so to speak, rather than including it in Q3 because I’m not really sure about the timing of those. Yes, Charles, I think you're referring to the step-up from Q3 to Q4 in our guide. So we're guiding about $1 billion of revenue in Q3 with a balance of just over $1.70 billion in Q4. Now that's slightly exaggerated the Q3 to Q4 step-up. When we originally looked at the top-down forecast for the second-half, there was a lower step-up in revenue mainly due to software increases in annual value and in software bookings that we expect. But when we looked at the hardware and the timing of when hardware deliveries happen, and there's so many that fall late in the quarter, we thought it was prudent to assume that those bookings, we should include them in our implied Q4 guidance, so to speak, rather than including it in Q3 because I’m not really sure about the timing of those.

Operator, Operator

We'll take our next question from Vivek Arya with Bank of America.

Unidentified Analyst, Analyst

Hi, this is Vivek. I wanted to start by asking how you measure the benefits from AI. In the semiconductor sector, several AI-related design initiatives began a few years ago. Have you seen these benefits reflected in your financials yet, or should we anticipate more rapid growth in the future? Additionally, it would be helpful if you could discuss the attach rate of your AI-enabled products to baseline contracts. Thank you.

Anirudh Devgan, President and CEO

Yes, AI has various implications for us. Firstly, we facilitate the development of AI systems as more of a horizontal application. For instance, our partnership with Nvidia and other companies like Tesla, which is working on both server-side and automotive AI, highlights this. We are uniquely positioned to supply components for AI applications across GPUs, data centers, and automotive sectors, and this trend is ongoing. There will continue to be more designed systems where our tools and intellectual property play a significant role. I believe we are still in the early phases of this expansion and monetization. Secondly, we are applying AI vertically to our own products. Our JedAI platform stands out as a data platform for AI, coupled with five unique AI applications covering various areas such as analog, digital, verification, packaging, and systems design. I feel we offer the most comprehensive product portfolio for chip package system design. Although we are still in the nascent stages, we are witnessing growing adoption of these vertical applications. We're also leveraging them for our internal chip designs for Palladium systems and have a substantial application engineering team collaborating with customers. So, we are not just providing solutions to customers and partners in EDA and SDA, but we are also gaining substantial internal benefits. Lastly, looking ahead, I believe AI will drive new applications, particularly in life sciences. We are laying the foundation for this with our OpenEye initiative, which presents significant opportunities for applying AI in digital biology over the next five to ten years. Altogether, we see a horizontal focus on building AI systems and GPUs, vertical applications in EDA and systems for our customers and ourselves, and emerging areas in life sciences. Overall, I feel we are well-positioned to take advantage of this major trend.

John Wall, Senior Vice President and CFO

Yes. To address your revenue question, it's evident that we're seeing quick results in the hardware area. On the software side, it may take a couple of contract cycles for changes to surface. We're experiencing strength in certain areas of software, but since it generates recurring revenue, it will take some time to fully materialize. However, the impact of AI is clearly evident in the strong demand for hardware.

Unidentified Analyst, Analyst

Got it. Helpful. And then just quickly as a follow-up. I hope we can discuss your capital allocation policy as well. Just given the pretty solid cash position and significant cash generation of the company, is there any interest in allocating more cash towards share repurchases? And also, if you could comment on any interest in issuing the dividend in the future as well?

John Wall, Senior Vice President and CFO

Yes, we constantly evaluate our cash allocation strategy. In the second quarter, we spent $325 million on stock buybacks, of which $200 million was from an accelerated share repurchase. The timing of this was intentionally aligned with our merit and promotion cycle. You may have noticed that our guidance indicates a rise in expenses and a slight decrease in margins in the latter half of the year compared to the first half. This is due to the salary increases and promotion cycle beginning on July 1 at Cadence. Typically, we also conduct our stock refresh for employees at this time. Since we grant a significant amount of stock during this period, our stock repurchase strategy focuses first on offsetting dilution and then gradually reducing the share count. We have discussed dividends with the Board on occasion, but we believe the best use of our excess cash is for stock buybacks. Our policy is to allocate at least 50% of free cash flow to repurchase shares, which should effectively counteract dilution, lead to a declining share count over time, and provide sufficient resources for business growth.

Unidentified Analyst, Analyst

Great. Thank you.

Operator, Operator

We'll take our next question from Jay Vleeschhouwer with Griffin Securities.

Jay Vleeschhouwer, Analyst

Thank you. Good evening. Business question first and a technology question as a follow-up. So Anirudh, as you know, for the last couple of years, we've often spoken about how semiconductor companies and systems customers can become increasingly alike. The question though is, in what important ways do they remain dissimilar in terms of sales cycles, conversion of RPO to revenue, services requirements, account profitability? Anything you care to mention that continues to distinguish those two classes of customers?

Anirudh Devgan, President and CEO

Yes, Jay, as you've noted, semiconductor companies are evolving into system companies, and vice versa. Many traditional semiconductor firms are now functioning more like system companies by integrating software, while a number of traditional system companies are venturing into semiconductors. There are notable similarities between the two. Regarding differences, one significant distinction is that system companies inherently combine hardware and software. A prime example is the automobile, which merges electronic and mechanical components along with hardware and software. Consequently, the strength we discussed in the hardware sector goes beyond chip design; it also encompasses software development. We are witnessing this especially in AI applications, where software must be developed concurrently with hardware. Even Nvidia, primarily known as a semiconductor company, has a substantial software component. Additionally, many data companies creating their own chips are heavily focused on software. Thus, the integration of software is crucial. Another aspect is the convergence of electrical and mechanical systems, which is why we are heavily investing in system design and analysis, including thermal simulation, electromagnetics, and computational fluid dynamics, particularly for data centers and cars. The unique factor is the combination of hardware with software and electrical with mechanical systems. In certain instances, we provide strategic services to help system companies initiate their projects, which is another key difference. We assist these companies in starting their chip development. In summary, these represent three significant differences. It's encouraging to see growth in the semiconductor sector as we embrace a new market with system companies, while we continue to value our semiconductor customers who are excelling and with whom we are collaborating in various capacities.

Jay Vleeschhouwer, Analyst

Okay. As the technology follow-up at DAC two weeks ago, there was a quite interesting presentation by Cadence at the conference related to AI, and there was the use of a very interesting with regard to productivity or where productivity is going to come from in future for EDA. And that term used by Cadence was task abstraction. So a new kind of abstraction from the decades of extraction we've seen in EDA for the last many years. So how are you manifesting that new concept of abstraction in terms of product development, packaging, all those sorts of things going forward since that term seems to be very much a critical part of your thinking going forward for EDA productivity?

Anirudh Devgan, President and CEO

Yes. If you take a step back and consider fundamental innovations, there are two significant developments in mathematics and computer science: abstraction and parallelism. In the history of electronic design automation, we have progressively moved up the levels of abstraction, from polygons to transistors, to gates, and finally to RTL. Over the last decade, the focus has shifted back towards parallelism, largely driven by advancements in cloud computing and CPUs. However, with AI, there is a real opportunity to achieve even greater abstraction and elevate our approach further. Essentially, the goal is to automate repetitive tasks that users typically handle, allowing the AI engine to take over. This increase in abstraction will ultimately lead to substantial automation and productivity gains. We see the potential to apply AI to various aspects, including design technology co-design with top foundries, as well as in activities such as floor planning and higher-level functions that were formerly done manually, like 3D-IC partitioning. Overall, the integration of AI to enhance abstraction and automate mundane tasks, thus allowing users to concentrate on more complex problems, is just the start. This applies across various areas, including digital verification and system design, among others.

Ruben Roy, Analyst

Thank you. Anirudh, if I could start from where you left off there, I had a question similar to Jay's. But I was wondering, I think you mentioned that SerDes with AI was deployed at eight of your top 10 customers. I'm just wondering today, can you talk about how AI is being used? Is it mostly placed in route? Or are you seeing some of these other areas being implemented in leading-edge designs, whether it's packaging or even some of the floor planning, etc.? How far along are we?

Anirudh Devgan, President and CEO

It's becoming increasingly widespread. Reflecting on the progress we've made over the past two to three years with SerDes and our clients, I'm confident that in a few years, almost all our customers will be utilizing these tools since this trend seems irreversible. Placement routing was one of the initial areas with significant adoption, but its influence is expanding to include foundries, floor planning, higher-level functions, and verification. I strongly believe that the most significant impact of AI in EDA will be in verification, as it remains an unbounded NP-complete problem. The need for ongoing verification is never complete, which is why we are making substantial investments in Verisium. We were the pioneers in applying AI to verification, and we have shared several customer examples in our prepared remarks. Moreover, in areas that historically lacked automation, like package and PCB design, Allegro X AI is making automation feasible for the first time in decades. System simulation in EDA has a three-decade history focused on optimization and automation, primarily through classical techniques. However, in fields such as CFD or electromagnetics, the ability to simulate has been limited. Now, thanks to our faster algorithms and GPU acceleration, we can perform high-fidelity simulations of entire systems, such as cars. Beyond just simulation, we can now optimize designs, like the shape of a wing, which ultimately meets user needs. This approach is highly relevant in the growing market of SD&A, where we're already seeing optimistic outcomes, such as optimizing wing shapes for racing cars using AI. Looking ahead, I believe these AI capabilities will be deployed across various platforms. We began with placement and routing, but we're witnessing promising potential beyond that. Ultimately, I think the greatest opportunity in EDA lies in verification, with a fair amount of optimism also surrounding system simulation and AI.

Ruben Roy, Analyst

Thanks very much, Anirudh for that. A quick follow-up for John. Just on operating margin. Down a little bit in Q3 with what sounds to me like hardware, kind of, strengthening towards the end of the quarter and obviously into Q4. How are you thinking about that? Or how should we think about operating margin and sort of the moving parts between Q3? You bumped it up a little bit for the full year. So if you can help us out on how you're thinking about operating margin in the context of hardware attach rates going up as you exit the year, that would be helpful?

John Wall, Senior Vice President and CFO

Yes, great question. The first half of the year had a non-GAAP operating margin of just under 42%. For the second half, we expect the margin to be between 41% and 42%. This slight decrease is consistent with trends we've seen in previous years, as it typically drops in the second half. The reason for this is our merit cycle on July 1, which includes pay raises and promotions. Since around 90% of our workforce consists of engineers, these pay increases significantly impact our expenses starting from the beginning of Q3. This increase in costs is affecting the comparison between Q3 and Q4 as well as the second half and the first half. I've been cautious with our Q3 guidance on the hardware side due to concerns that some hardware deliveries scheduled for late September might slip into Q4, particularly considering the holiday period in China and other parts of Asia. Therefore, I have anticipated that most of those delays will affect Q4 in our guidance. This is influencing the operating margin we are showing for Q3, but overall, the second half should generally reflect a margin of 41% to 42% when accounting for that caution.

Ruben Roy, Analyst

Got it. Makes sense. Thank you, John.

Operator, Operator

We'll take our next question from Blair Abernethy with Rosenblatt Securities.

Blair Abernethy, Analyst

Thank you, gentlemen, and excellent quarter. I wanted to focus a bit on China. The year-over-year growth this quarter was quite impressive. Could you provide some insights on what is fueling that growth and how it might progress? Additionally, regarding hardware, John, you previously mentioned an increase in your production capacity. I’m curious about your current status halfway through the year and how your lead times are shaping up.

Anirudh Devgan, President and CEO

Yes, I'll begin, and then John can provide insights on the inventory. Overall, we are pleased to see ongoing strong design activity across various end markets in China, along with solid performance in the hardware sector. Specifically in Q2, there was significant strength from the mobile segment, with many large system companies developing their own chips, as well as a particularly robust performance in the electric vehicle automotive sector. China, much like the rest of the world, is a diverse market, and we are active in all its segments. However, those two areas showed notable strength in China. John, would you like to add?

John Wall, Senior Vice President and CFO

Yes. In response to your question about lead times, at the beginning of the year, we increased our production capacity because lead times were over 26 weeks. We realized this could negatively impact our competitiveness in the hardware market. In the first quarter, we made significant progress by ramping up production. We maintained those production levels in the second quarter and plan to sustain them for the rest of the year. Our aim is to reduce lead times to between 10 and 13 weeks, although we haven't achieved that yet. Based on my guidance for the third quarter, we don't need any additional Q3 hardware bookings to meet our targets. Currently, our backlog puts us slightly over the 13-week mark, likely around 15 weeks, so we won't complete all hardware orders in Q3. However, demand remains strong, and while we are reducing lead times, we’re still around the 15-week mark, and I aim to get that down to 10 to 13 weeks.

Blair Abernethy, Analyst

Okay. Great. And actually, just one more question on the hardware, if I can. So the Palladium Z2 and the Protium X2 have been out in the market now for a bit of time. When can we expect to see next-generation versions?

Anirudh Devgan, President and CEO

It's been only two years since the last system was launched, which was five or six years ago. We're quite pleased with how our current systems are performing. Even with the launches of Z2, X2, Z1, and X1, we're seeing great results. Additionally, we're continuously enhancing the software, which complements the hardware. There's an ongoing effort to improve software and compile times, independent of the hardware advancements. Overall, the team's performance has been exceptional, and we truly appreciate the support from our customers and partners.

Blair Abernethy, Analyst

Great. Thanks very much, guys.

Operator, Operator

We'll take our next question from Joe Vruwink with Baird.

Joe Vruwink, Analyst

Hi there. I have a question regarding your forecast for second half revenues growing by 15%. Some multiyear compound annual growth rates are around 15% to 16%. Looking ahead to the next few quarters, not just reflecting on past performance, can you discuss how your customers are behaving and what Cadence is planning to offer that might surpass previous experiences in terms of research and development focus and the potential share of their budgets? Instead of viewing this 15% growth as the peak of historical growth rates, do you think we could be witnessing a shift in the overall expectations for growth rates and what the company might achieve in the future?

Anirudh Devgan, President and CEO

Well, first of all, I would say a 15% growth rate and 19% EPS growth is an amazing performance of any company. So we are very proud of the team to deliver this. And in terms of design activity and R&D intensity, I mean that is strong, if not getting stronger. So of course, the overall macro environment as tough as it has been last year and this year, but we do see a lot of semi and system activity. So that's not any different than before, but we have to carefully monitor how things go in the future, of course. But overall, we are pretty pleased to see the design activity.

John Wall, Senior Vice President and CFO

Yes, Joe, in the CFO commentary we published, you will find an emphasis on tracking three-year compound annual growth rates. Our goal is to grow accounts through a full contract cycle, as most of our customers enter into three-year baseline contracts which allow for add-ons throughout the year that align with those baseline contracts. We believe the three-year CAGR is a relevant measure. Currently, if you examine the revenue growth rate on a three-year CAGR basis, it is continuing to decline. The rate of acceleration has certainly slowed over the past two years, hovering around the mid-teen mark. However, we remain focused on driving profitable revenue growth, and we generate substantial cash flow. We intend to use over 50% of that cash flow to repurchase shares and decrease the share count, which aims to increase our non-GAAP earnings per share. Our stock-based compensation has consistently been about 8% of revenue, and we have maintained a non-GAAP EPS CAGR of 20% or higher for the past five years. With the current midpoint guidance showing a 19% rate, we are on track for a 22% EPS growth on a three-year CAGR basis for 2023. Our main focus is on the bottom line; we prioritize EPS growth for our investors and improving profitability, with revenue growth serving as a means to that end. We are very pleased that this growth is continuing to increase.

Joe Vruwink, Analyst

Thank you for that. I have a question about AI that follows up on Ruben's inquiry. It seems that maturity, particularly regarding customer experience and referenceability, is strongest in the back-end area of Cerebrus. Considering the pace of adoption for the innovations you've launched since last fall, do you think it's faster or slower than what you've experienced with Cerebrus? Is there any reason that debone verification wouldn't be a prime candidate for your underlying technology? I'm curious if you could compare and contrast the adoption curve.

Anirudh Devgan, President and CEO

Yes, that's a good question. I would say that the adoption curves are quite similar at this stage, particularly with Verisium, where verification is a significant focus. They are somewhat easier to adopt given that customers have seen a proof point with Cerebrus, which has built their confidence in our ability to deliver comprehensive AI solutions. The same can be said for Allegro X and Virtuoso Studio. However, as is common in this business, customers need to experiment with the products on specific designs before moving to broader adoption. This pattern is present in new areas, including Verisium and Allegro. Overall, I would say the pace of adoption is similar, and there is considerable interest in other AI solutions beyond just Cerebrus. Although the adoption rate is comparable at this time, it will take some time, which is perfectly fine. It’s a natural process of testing and implementing. In some areas, like hardware, the pace may be quicker due to how revenue recognition works. But overall, adoption is progressing well.

John Marco Conti, Analyst

Hi, Anirudh and John. Thanks for taking my questions. And so the first one is, I know you've touched on this a little bit. Could you share some more color on the usage of AI tools in system companies this quarter? Are any of the new wins related to AI tools for system customers? And how can we think about the acceleration of usage, not just leading nodes but in the broader design space, even at more mature nodes? I think investors are thinking what next when you taking the current use cases in the next two, three years, where the cost trade-off might not be as imperative as to say in the GPU market? I imagine is this a function of better price recovery matched with higher adoption/proven use cases.

Anirudh Devgan, President and CEO

Yes. I believe the application of AI has both broad and specific aspects. The broad aspect involves developing the AI infrastructure, which primarily focuses on advanced node GPUs and system companies. One of the notable examples is Tesla, which has a server-side chip known as Dojo, as well as capabilities in its vehicles for full self-driving. This is a significant use of AI that will likely expand as more automotive companies begin to implement similar technologies. Additionally, we have several hyperscaler customers that are creating their own silicon for AI, which is widely known. We are collaborating with them, and we saw several significant expansions in the second quarter related to AI chips from these large hyperscale firms. On the specific side, our five main product portfolios for AI, including those built on JedAI, apply to all nodes and are not limited by specific technology nodes. The initial focus for general AI applications started at advanced nodes, but as Edge computing becomes more prevalent, it will also extend to more mainstream nodes since it does not require cutting-edge technologies. Particularly, vertical applications of AI in tools like Virtuoso or Allegro can be utilized across a range of nodes, not just advanced nodes, and the benefits we observe are consistent regardless of the node. For instance, depending on the design, we sometimes achieve a 10% to 15% improvement in power, performance, and area from AI. Transitioning from one node to another can yield similar benefits of about 10% to 15%, indicating that better AI algorithms can provide advantages comparable to moving to a new node. Therefore, this also enhances efficiency at the current node, and applying AI to EDA and system design and analysis will be independent of the node used. Yes, that's a good observation. One trend driving full flow adoption is the move from seven to five to three to two advanced nodes. This transition complicates things, and we designed our system years ago to be fully compatible with this approach. Additionally, AI orchestration enhances performance, as the AI engine can optimize all aspects of the flow. This connection is advantageous for us. This advancement is significant and is also evident in other areas like verification. Both these trends are speeding up the deployment of full flow, advanced nodes, and AI.

Joshua Tilton, Analyst

Hey, guys. Thanks for sneaking my two questions at here. The first one, I want to go back to the full-year guide. I know that you guys mentioned that you expect, I think, hardware to grow 15% year-over-year in the second-half. But what about compared to the first half? Are you still baking in any conservatism in the hardware guide relative to the strength in hardware that you saw in the first-half of the year?

John Wall, Senior Vice President and CFO

So Josh, just to correct that, the expectation for 15% growth in the second-half is comparing second-half versus second half or total revenue at Cadence, all our businesses. We expect all of our businesses to grow by double digits in the second-half, compared to the second half last year. Some are stronger than others. In relation to hardware specifically, we've seen strong demand pretty consistently for a couple of years now. In Q1, we ramped up production capacity because lead times had gone over six months. We tried to address those, and it certainly helped with improved demand in the second quarter, and we're seeing more opportunities in the pipeline for the second half. Now that gives us comfort to raise the guide for the second half of the year. But it is a pipeline business. Typically, people only put their opportunities into the pipeline about a quarter before they intend to purchase. In some cases, you might get visibility four to six months in, but it's a pipeline business. Right now, we have strong backlog for hardware, and that's spread into our decision to increase our guide for the second half, but there is potential for that demand to continue. And if we're going to keep production capacity at the levels we're at right now because we expect that demand to continue into the future.

Operator, Operator

I'll now turn the call back over to Anirudh Devgan for closing remarks.

Anirudh Devgan, President and CEO

Yes. Thank you all for joining us this afternoon. It's an exciting time for Cadence with strong business momentum and growing opportunities in the semiconductor and systems industry. With a world-class employee base, we continue to deliver on our innovation road map and delight our customers and partners. On behalf of our Board of Directors, we thank our customers, partners, and investors for their continued trust and confidence in Cadence.

Operator, Operator

Thank you for participating in today's Cadence Second Quarter 2023 Earnings Conference Call. This concludes today's call, and you may now disconnect.