Cadence Design Systems, Inc. (CDNS) Q2 2025 Earnings Report, Transcript and Summary

Ladies and gentlemen, good afternoon. My name is Abby, and I'll be your conference operator today. At this time, I would like to welcome everyone to the Cadence Second Quarter 2025 Earnings Conference Call. [Operator Instructions] Thank you. And I will now turn the call over to Richard Gu, Vice President of Investor Relations for Cadence. Please go ahead.

Thank you, operator. I'd like to welcome everyone to our second quarter of 2025 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. The webcast of this call and a copy of today's prepared remarks will be available on our website, cadence.com. Today's discussion will contain forward-looking statements, including our outlook on future business and operating results as well as the impact of our DOJ and BIS settlements. 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, all financial measures discussed on this call are non-GAAP unless otherwise specified. The non-GAAP measures 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, I would ask that you observe a limit of one question only. If time permits, you can requeue with additional questions. Now I'll turn the call over to Anirudh.

Thank you, Richard. Good afternoon, everyone, and thank you for joining us today. Cadence delivered exceptional financial results for the second quarter of 2025, exceeding our Q2 revenue and EPS guidance driven by ongoing broad-based strength across our AI- driven product portfolio. Bookings were stronger than expected, highlighting the strategic relevance of our AI-driven portfolio and the depth of our customer relationships. Demand for our technologies continues to grow, driven by customers embracing our products at scale, and we are raising our financial outlook for the year to 13% revenue growth and 16% EPS growth for 2025. John will provide more details on both our Q2 results and the updated outlook. We continue executing to our intelligent system design strategy initiated in 2018, which remains a clear differentiator in a rapidly evolving landscape. Our early investments delivering to our vision of unified EDA, IP, 3D-IC, PCB and system analysis are paying off. These capabilities are enabling us to lead through the accelerating waves of the AI super cycle from AI infrastructure build-out to physical AI in autonomous systems to the emerging frontier of sciences AI. Customer R&D investments remain robust, particularly as AI drives exponential design complexity, such as in advanced node design and complex system architectures. And this is translating into broad-based demand across our portfolio. Embedding Agentic AI into our design platforms across core EDA, system design and system simulation workflows enables the evolution from a traditional tool-based flows to autonomous goal-driven agents. Our Cadence AI portfolio powered by multiple autonomous silicon agents and built on our unified JedAI platform, with NVIDIA accelerated compute is delivering optimized designs and massive efficiency gains for our customers. At CadenceLIVE 2025, we introduced the new Millennium M2000 AI supercomputer featuring NVIDIA Blackwell, delivering AI accelerated simulation at unprecedented speed and scale across engineering and science workloads. This tightly co-optimized hardware/software full system stack delivers up to 80x higher performance and up to 20x lower power versus traditional CPU-based systems. Multiple customers provided endorsements, including Ascendance, MediaTek and Treeline Biosciences. In Q2, we furthered our long-standing partnership with ADI through a broad proliferation of our core EDA software, including AI- driven Cadence Cerebrus and Verisium solutions as well as system software across PCB, advanced packaging and system analysis. Also in Q2, we deepened our partnership with SK Hynix through a broad expansion of our EDA software, system software and design IP solutions. And a major semiconductor company meaningfully expanded its relationship with Cadence in Q2 through a broad proliferation of our EDA, IP and SDA portfolio. We furthered our long-standing collaboration with TSMC to accelerate time to silicon for customer designs using 3D-IC and advanced node technologies such as TSMC's A16 and N2P through certified design flows, silicon-proven IP and ongoing technology collaboration. We continued the strong momentum in our IP business, delivering more than 25% year-over-year growth in Q2. driven by product strength and a broadening silicon solutions portfolio. AI and HPC use cases spearheaded the strong demand for our IP offerings with advanced technologies such as HBM4 and 224-gig SerDes, notching key wins for scale-up and scale-out in the AI infrastructure space. We built on our strategic collaboration with emerging advanced foundry as they awarded us a large deal in Q2 for our leading HBM4 solution. We introduced the industry's first LPDDR6 memory IP offering up to 50% higher performance to meet the growing memory and capacity needs of AI LLMs and Agentic AI workloads. At CadenceLIVE 2025, we launched the Cadence Tensilica NeuroEdge 130 AI Co-Processor to accelerate physical AI applications. And in Q2, a market-shaping wireless technology company selected Tensilica HiFi 5s as the standardized audio solution for its music and voice platforms. Our core EDA revenue grew 16% year-over-year in Q2. Further proliferation of our digital full flow at the most advanced nodes continued and more than 50% of advanced nodes designs using our implementation solutions are now using Cadence Cerebrus. In Q2, we launched Cadence Cerebrus AI Studio, the industry's first Agentic AI, multi-block and multiuser SoC design platform. This technology delivering up to 20% PPA improvement while accelerating chip delivery time by 5x to 10x was endorsed by Samsung and STMicroelectronics at launch. And Renesas, successfully used our Pegasus physical verification solution to sign off an advanced node SoC after it demonstrated a significant throughput advantage. Our industry-leading Palladium Z3 and Protium X3 platforms accelerated their momentum, delivering outstanding results with Q2 being the best revenue quarter ever for our hardware systems. Demand for hardware was strong and broad-based, driven by AI, HPC and automotive customers. Our verification software suite that includes Verisium, Xcelium and Jasper and leverages big data and AI to optimize verification workloads saw continued expansion with 27 new logos in Q2. Building upon 30 years of industry leadership, we launched the Virtuoso Studio 25.1 release, offering broad support for RF, photonics, mixed signal and advanced heterogeneous designs. Our leading Spectre X circuit simulator closed several deals with strong growth, while our FastSPICE circuit FX platform has now been adopted by the top 3 memory companies. Our system design and analysis business delivered another standout quarter with 35% year-over-year revenue growth. On the packaging front, there was strong customer uptake of our 3D-IC technology and top foundries and semi customers embraced our AI- driven advanced substrate router, which provides tremendous productivity benefits. Our AI-driven Allegro X PCB design platform saw continued proliferation as multiple aerospace and defense, hyperscale and EV customers took advantage of the platform's meaningful productivity and next-generation capabilities. Our Clarity and Celsius solvers saw significant expansion at a major hyperscaler and Clarity secured a key win at a marquee AI company, while our reality data center digital twin drove strong growth at a top hyperscaler. BETA CAE technology integrations with our CFD, thermal and electromagnetics products were released as BETA CAE solutions continue to score key competitive wins, particularly in the automotive segment. Finally, I'm pleased to share that we have entered into a settlement with the U.S. Department of Justice and the U.S. Department of Commerce's Bureau of Industry and Security that resolved the previously disclosed investigations into certain transactions with customers in China that occurred between 2015 and 2021. The settlement represents a mutually acceptable path forward for all parties, and we believe are in the best interest of our customers, partners and shareholders. I want to emphasize that Cadence is deeply committed to the highest standards of compliance, and we have significantly enhanced our compliance processes over the last few years and continue to implement improvement measures to proactively address evolving trade restrictions. We remain focused on delivering for our customers and shareholders and executing the clear strategy we have laid out to drive innovation and enhanced value creation. In summary, I'm delighted with our Q2 results and the continued momentum across our broad and innovative portfolio. The AI- driven era presents tremendous opportunity and the co-optimization of our comprehensive EDA and SDA portfolio with accelerated computing and Gen AI uniquely positions us to deliver breakthrough solutions across a wide range of markets. Now I will turn it over to John to provide more details on the Q2 results and our updated 2025 outlook.

Thanks, Anirudh, and good afternoon, everyone. I'm pleased to report that Cadence delivered excellent results for the second quarter of 2025 with broad-based momentum across all of our businesses. Strength in other regions more than offset the impact of the export restrictions on China outlined in the BIS letter dated May 23, which was later rescinded. Robust design activity and customer demand, coupled with our strong execution, drove 20% revenue growth and 29% non-GAAP EPS growth year-over-year for Q2. Here are some of the financial highlights from the second quarter, starting with the P&L. Total revenue was $1.275 billion. GAAP operating margin was 19% and non-GAAP operating margin was 42.8% and GAAP EPS was $0.59 with non-GAAP EPS $1.65. Next, turning to the balance sheet and cash flow. Cash balance at quarter end was $2.823 billion, while the principal value of debt outstanding was $2.5 billion. Operating cash flow was $378 million. DSOs were 51 days, and we used $175 million to repurchase Cadence shares. Before I provide our updated outlook, I'd like to share what's embedded. As Anirudh mentioned, I'm pleased that we've reached a settlement with the DOJ and BIS, resolving previously disclosed investigations into certain China sales from 2015 to 2021, totaling approximately $45 million over the 6-year period. As part of the agreements, we will make a payment of approximately $141 million in our third fiscal quarter. Please see our Form 8-K, which includes additional details regarding the terms of the agreements. On July 4, 2025, the One Big Beautiful Bill Act was enacted in the United States. This act includes the restoration of favorable tax treatment for certain business provisions, including the immediate expensing of United States research and development expenditures. We expect it to decrease Cadence's United States federal tax payments for the remainder of fiscal 2025 by approximately $140 million. Our updated outlook includes the timing of the settlement penalty, the cash tax benefit of the OBBBA and the usual assumption that export control regulations that exist today remain substantially similar for the remainder of the year. Our updated outlook for 2025 is revenue in the range of $5.21 billion to $5.27 billion, GAAP operating margin in the range of 28.5% to 29.5%; non-GAAP operating margin in the range of 43.5% to 44.5% GAAP EPS in the range of $3.97 to $4.07. Non-GAAP EPS in the range of $6.85 to $6.95. Operating cash flow in the range of $1.65 billion to $1.75 billion, and we expect to use at least 50% of our annual free cash flow to repurchase Cadence shares. With that in mind, for Q3, we expect revenue in the range of $1.305 billion to $1.335 billion, GAAP operating margin in the range of 32% to 33% non-GAAP operating margin in the range of 45% to 46% GAAP EPS in the range of $1.14 to $1.20 and non-GAAP EPS in the range of $1.75 to $1.81. 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 and GAAP to non-GAAP reconciliations. In conclusion, I'm pleased with our strong first half results and the robust pipeline for the second half of the year. At the midpoint, we now expect revenue growth of 13% and non-GAAP operating margin of 44% for the year. 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 Instructions] As a courtesy to all participants, we ask that you limit yourself to one question [Operator Instructions] And our first question comes from the line of Joseph Vruwink with Baird.

I wanted to ask a question on physical AI. It seems like over the past quarter or so, many of your key development partners have had more to say around what they're doing with edge devices or even small language models maybe as a means to enabling physical AI. Is this factoring into the bookings strength you've seen recently? And is it maybe leading to more spend or different spend with Cadence just in terms of the tools that this is going to need versus what the initial build-out of AI infrastructure has meant?

Yes, Joe, it's a great question. And first of all, I'm very pleased by our results and our performance and the demand of our products, which is broad-based. And also, I think there is -- first of all, I believe there is overall optimism in the benefits of AI in our customers, both from what they can -- their own products and also how they can use AI internally. So therefore, they are investing more in their innovation and given the critical nature of our products, investing more in Cadence. Now it has several aspects to it. And I have been a big fan of physical AI for a long time because one unique advantage we have in Cadence is the privilege to work with all the top companies in the world. And we believe that, of course, AI infrastructure is huge, but physical AI has the potential of being even bigger and then follow that with sciences AI. That's why we have laid this 3-phase evolution of AI. And now if you look in the marketplace also with autonomous cars or robots and drones, it is becoming much more public. And our advantage is, even though some of these things come out later, the customers start investing in R&D before they come out in public. So -- but I think physical AI will play a very key role for our products because the silicon required, first of all, and physical AI will affect the whole three layers of that AI cake that I've talked about. So first of all, the silicon is different in the car or in the robot or in the edge devices is different than data center silicon. I mean it's still AI-driven, but it's more power optimized, runs on lower battery, as you know. So the silicon is different. The simulation and design is different. And of course, the AI models themselves are different. They are more word model than LLMs. But all these physical AIs still need to be trained. Even if the inference like for autonomous car runs on the car, the actual AI model is trained on the data center. So the beautiful thing of physical AI is not only it creates new opportunities for us, it also emphasizes the importance of AI infrastructure in the data centers. So it is helping both sides of that equation. And so we are benefiting from that. And we are, as you know, working with all the main AI data center players as they design chips and systems. So the impact is both on the data center side and the edge side. But there's still an evolving market. I think physical AI is still in the early innings. There's still like three to five years of more development to go. So -- but overall, I think what I would like to say is that the customer environment is, I feel personally is better than it was six months ago.

And our next question comes from the line of Gianmarco Conti with Deutsche Bank.

I mean, firstly, congrats on another amazing quarter. Simply, what led to Cadence increase in the growth outlook, even though you could not recognize one month of China revenue? I guess the curiosity is whether there was a single stack of renewals across EDA or was this across all fronts? And maybe you can give us more comment on backlog and the development throughout the year.

Yes, Gianmarco, great question. I mean, yes, it's been an interesting quarter. I mean China was -- ended up being 9% of our revenue in Q2. That's down from 11% in Q1. But we've seen strong demand across all geographies. And strength in other regions more than offset any near-term softness related to China during Q2. We've spoken in the past about how well diversified our customer base is. And we're increasingly seeing growth, and we're seeing the growth in bookings from AI, HPC and system design workloads globally. But we're very, very pleased with the way backlog ended up at the end of Q2. It is stronger than we expected going into the quarter despite all of the restrictions. But -- and yes, we're very, very pleased with where we are halfway through the year. Anirudh, anything to add?

No, John, that's right. I mean, overall, I would like to say the demand is broad-based. You can see it in all the results of all the 3 main lines of business. I mean hardware is doing phenomenally well. We had a record quarter ever in terms of revenue. and we have a clear lead in hardware. And also, we are essential to all the major AI chips being designed using Palladium and our EDA software. And then all these agentic AI tools like Cerebrus AI Studio, I mean that's a phenomenal new product and then Verisium, Allegro X. So I think both the software and hardware business is doing well in core EDA. And then IP had a great quarter. I mean there's a lot of reasons behind that. One is the AI infrastructure build-out, but also there are at least four major companies doing advanced node foundries now with TSMC, our long-standing partner, Samsung, even today, there's a big announcement from Samsung Foundry, Intel with 18A, 14A and Rapidus in Japan. I just came back from Japan with this big opening of Rapidus. So there are at least four advanced node foundries that all require IP. So I think that's also driving strength in IP. And then system continues to do well because of our focus on 3D-IC, which is the fastest-growing part of the system market. And beta is providing us a good kind of integration with rest of the flow and new products like Millennium. So if I look at all the three main areas, I think I feel we are very well positioned and the market itself seems to be improving with the AI super cycle.

And our next question comes from the line of Vivek Arya with Bank of America.

Just a near and a longer-term China impact question. So on the near term, how much of a headwind was China in Q2? I know, John, you mentioned they went from 11% to 9%, but what was kind of the expectation? Then if we zoom out for all of 2025, I think in the past, you had said China sales were expected to be flat year-on-year. Is that still the right approach because that would still imply quite a bit of a lift in the back half? And then Anirudh, if we look longer term, what is the right China exposure for cadence? Does it naturally just come down over time? Or will it probably stay at this 9%, 10%, 11% kind of range over the longer term?

Yes, Vivek, look, I'll start because I understand your question. I mean I would view our outlook for China to be optimistic, but prudent. The -- I mean, our guidance reflects what we believe to be a prudent and well-calibrated view of the second half of the year. The export control environment is dynamic. And while we've incorporated current regulatory framework as of today into our assumptions, we always add some prudence to account for potential variability, whether that's geopolitical or operational. But we're very, very pleased with how China is doing. It's -- I know last quarter, we told you that we were expecting it to be flat. It's hard to see how China won't increase a little bit over last year, but we've been prudent with our guide.

And Vivek, long term, I think China will -- of course, is going to invest in chip design and system design, just like all geographies. But I think the percentage of revenue should be similar or maybe a little down, but because -- not because China won't do well, but I think the rest of the world is doing phenomenally well, right? All the investment you're seeing in U.S. and then Japan, Korea. I mean, so not to say in particular about China, but I think the rest of the -- and which we saw in Q2 also, there is significant investment. So given that context, it's difficult to predict exactly what China will do. But it's good to see that China is doing well, but the rest of the world is doing even better.

And our next question comes from the line of Harlan Sur with JPMorgan.

Great job on the quarterly execution. If I look at many of the AI xPU, ASIC and merchant chip design programs that are in design right now, many of them are looking to transition from 2.5D to 3.5D advanced packaging architectures, which includes chip stacking, right? And many of these programs are going to start taping out second half of this year. And in some cases, your customers are integrating up to like 10 chips in a single package, right? This is a very complex undertaking, integration, floor planning on top of that, you got signal integrity, thermal power challenges. Wondering how much is this contributing to the bookings and revenue strength as more of your customers are adopting your Integrity 3D-IC or your Allegro X advanced packaging platforms to tackle these challenges of 3.5D packaging? And then how much is advanced packaging roughly contributing to your overall revenues?

Yes, Harlan, that's a great question and a great observation, of course. I mean the whole industry, especially in HPC and AI is moving to this chiplet-based architectures. And also, I think it's not just limited to the data center. Even if you look at the latest auto designs and all -- all the other markets will, I think, over time, move to this new packaging architecture. And we are -- Cadence is uniquely and very well positioned. I mean I think we have talked even earlier, Allegro is the platform of choice for package design. And 3D-IC is another way of talking about package design. And then at the same time, we have Virtuoso, which is analog, Innovus, which is digital and then all the system analysis tools like Clarity and Voltus and Celsius for. So -- and that is all incorporated into Integrity. And then we closely worked with TSMC. TSMC had done a fabulous job, by the way, in 3D- IC, and we have worked closely with TSMC over the last several years to develop this 3D-IC flow that is used by most of their main customers. And then now Rapidus and Samsung and Intel, we're working with all the other foundries to develop this kind of 3D-IC flow because it will be critical for all the other foundries. And we don't explicitly call out Allegro in our SDA business, but it is a significant part of that business, but also it pulls in the other things. It's not just Allegro by itself, but it naturally pulls in analysis tools and clarity and all those things and even the base tools like Virtuoso and Innovus. But it is a platform of choice for all the major companies as they implement this new 3D-IC or now 3.5D IC technologies. And this is only in the beginning. I think even in TSMC OIP, they showed road map that this is only going to increase. I mean this is an orthogonal access to -- I mean, you know this, this is orthogonal access to Moore's Law. So Moore's Law, first of all, okay, we are always worried about the natural questions from investors or employees sometimes, how long will the Moore's Law continue? So first of all, Moore's Law anyway is going to go to at least 1 nanometer, right? So we are at 3, 2, 1.41, okay, that's 10 years. And I visited some of our research partners like Imec and they're planning Moore's Law to go until 2042 with new transistor structures. But at least for the next 10 years, I see Moore's Law being strong. But then this 3D-IC and heterogeneous integration provides orthogonal levels of integration. And if you look at TSMC and other road maps, right, they have very aggressive road maps to be able to put more and more chips, like you said, in a package. So we are pushing on both of these dimensions. Moore's Law, we want to make sure we are aligned with all the latest technologies and customers and then this 3D-IC and heterogeneous integration.

Our next question comes from the line of Lee Simpson with Morgan Stanley.

Well done on another great quarter. I think it falls to me to maybe ask about the Agentic systems. I know again, this is the second quarter you brought it up. It does look as though development is moving ahead. And I think if the comments are to be interpreted right, you are seeing some early sales, one assumes in sort of pilot line development. But I'm trying to -- I'm still trying to put this into perspective. What -- if we take a step back, do we need a new business model or a different go-to-market strategy to get full value here? And more generally, how will you monetize this added value that an Agentic system will bring to the customer? Just any thoughts around that and maybe timing as well because it does look as though this is relying on still early-stage reasoning models.

Yes, Lee, that's a good question. So, I mean, as you know, we package them separate from our base tools. So of course, base tools are phenomenal. But then we have these Agentic workflows on top of our base tools. And customers are embracing both our base tools and the Agentic AI flow. I mean two great examples. I mean one of them I mentioned briefly is in the back end, Cerebrus. Cerebrus by itself, like I mentioned, is more than 50% of our designs are already using Cerebrus which, let's call it classical AI. But now with Cerebrus AI Studio, it's a whole workflow. So it's more -- it's an agentic AI solution instead of just doing block implementation, it does floor planning, it does timing closure. So what typically a designer could do like 3 million to 5 million instance design, they could do like 30 million to 50 million instance. So it's a massive productivity and PPA benefit as the AI does more of the manual work that was manual in the past. So that tool itself had a lot of early adopters like we mentioned on the call, Samsung and ST and others. And then there is on the other side, which is verification and RTL writing. This whole notion of LLMs generating and reasoning element generating code is a big thing, not just in software development, like CC++ but also chip design and RTL. So those two areas are very, very positive. One is in the front end with RTL generation and verification and the other is in the back end in PPA optimization. And that -- those are different tools than our traditional tool set, and we engage with customers on that. And our philosophy always is because we have a long history of innovation and automation in EDA, our philosophy is to deliver value to customers. Their workload is going up anyway. and align with the top customers. And usually, they will reward us for that. And that's our history over the last 10 years. And so we are focused on innovation and productivity. And we have all kinds of business models to monetize that in any way, and we'll see how that progresses over time.

And our next question comes from the line of Jim Schneider with Goldman Sachs.

I was wondering if you maybe could talk a little bit more about the core EDA results, very strong in the quarter with a lot of growth. Can you maybe cite some of the drivers of the strength in the quarter, be it new customers or Cerebrus pricing benefits or anything else that was onetime in nature? And maybe give us a sense about how you expect the Cadence of the core EDA revenue to trend in the back half of the year.

EDA -- core EDA is doing phenomenally well. I mean just to remind -- I mean, I think most of our investors know this already, but just to remind that we have the broadest portfolio in core EDA. We have digital, which we are leading position in, especially in the TSMC ecosystem, Virtuoso, which is de facto standard in analog mixed signal. Verification, we have all the verification software tools and Palladium and Protium and hardware. So Cadence has the most comprehensive EDA portfolio on the market. And as AI adoption happens, it's both the core product portfolio plus the AI-driven agents that we talked about, and we saw signs of that in Q2. And then some of the key customer wins we highlighted is, of course, SK Hynix, they're doing phenomenally well, as you know, with AI and HBM. ADI, which is a long-term Cadence partner. And then overall strength in hardware, which was very broad-based, then in IP and systems, which is outside of EDA. So overall, I think I'm pleased with that. Of course, as you know, we never focus on one individual quarter. There could be quarter-by- quarter variation. But overall, I think EDA is doing well, and I expect it to grow going forward.

Yes. And Jim, we're getting proliferation at marquee customers, and we're seeing the second half looks particularly strong on the software side as well as hardware on the -- in core EDA.

And our next question comes from the line of Jason Celino with KeyBanc Capital Markets.

John, if I think I heard you correctly, I think you said that China would be up a little bit this year versus flat previously. This is on top of, I assume, the China restrictions that were temporary. So this in itself seems important. I don't know if you'll be able to indulge us a little bit, but what do you think China growth could have been if those restrictions never happened, like if we never had those 6 weeks?

Yes, Jason, I mean, great question. Very, very difficult to kind of figure out what revenue would have been in that kind of parallel universe where that never happened. The one thing I take comfort from, though, is that the restrictions came and they went. But -- so I think the focus on the year. And when I look at the year that previously, we thought the year would be flat for China with the strength that we've seen across the board, across all businesses and across all geographies, it's really hard to see China remaining flat year-over- year now that I think it will be slightly up. But of course, we're normally very prudent with our guide, and I thought it was appropriate to remain prudent with the outlook for the year. So we've been cautious but optimistic with that outlook.

And our next question comes from the line of Gary Mobley with Loop Capital.

John, when we entered the year, I think your expectation, correct me if I'm wrong, was the year with a strong renewal period in the second half. And clearly, your bookings in the first half of the year have exceeded your expectation by, I assume, several hundred million dollars. And so my questions are two part. I want to confirm that the June quarter ending backlog excludes China. And with the strength in the first half that you've seen, what does that tell you about the potential for the second half bookings and exiting the year with perhaps record levels of backlog?

Yes, Gary, I mean, very astute question. Yes, we -- to confirm, we had to exclude a number of China bookings from our backlog by the end of Q2, we had to reserve for those because at the end of Q2, the restrictions were still in place for us. So the closing backlog at the end of Q2 reflects a lower level of backlog than it would have been had those China restrictions been rescinded prior to June 30. But -- and then in terms of the outlook for the year, yes, I'm pretty confident we're going to end up the year with a higher backlog than we started the year. So I'm very comfortable that we'll end up with a book-to-bill of 1. Second half bookings, the renewal cycle is strong in Q3 and Q4. I think both Q3 and Q4 will have bookings that exceed our revenue in those quarters. But -- and like I say, we should -- we expect that at the end of the year, we'll have a new higher and record level of backlog than we had last year.

And our next question comes from the line of Jay Vleeschhouwer with Griffin Securities.

Anirudh, you spoke earlier in answer to an earlier question and in your prepared remarks about Agentic AI. And I'd like to ask about the broader implications and requirements from that. One of the terms that's come up this year more broadly in software, not just in EDA, having to do with Agentic is orchestration. And in your world, specifically, if we think about what you're providing with Agentic AI or AI generally, it's fundamentally, I think, a form of simulation. And therefore, the requirements for that would also seem to be new forms of process or data management and traceability, which is a critical function in simulation, if our thesis is right about what you're really doing. So beyond just introducing these agents and aids, how are you thinking about the broader portfolio and capabilities that you need to provide customers, particularly since you refer to their workflows?

Yes, Jay, that's a great point. So yes, you're absolutely right. I mean we want to make more of a workflow automation, just like I mentioned with Cerebrus AI Studio. So it's not doing a point function. It is doing multiple functions together with reasoning. And the critical need is apart from the LLMs and all, there's a critical need for like a data structure or database to store all these actions. So what -- I'm pretty pleased about is the response of our customers to JedAI. We talked about JedAI being our joint enterprise data and AI platform. And it has both the data storage to -- because we need to capture not just one tool or one point in time, multiple tools and multiple flows like just like a human would do. So JedAI has become a very essential part of our AI deployment to customers. And it's a very flexible system because some of our customers, some really big customers want JedAI to be on-prem because their data is very, very sensitive. Some customers are okay with JedAI being on the cloud, okay to use cloud LLMs or cloud data management. And then some customers want a hybrid on-prem and cloud solution. So JedAI uniquely positions us to make that kind of invisible to the user. So -- but JedAI is critical along with the AI agents to deliver this solution to our customers. And we are able to do much more, and we'll do much more of a full workflow solution along with JedAI and then the agents on top, whether it's Verisium or Cerebrus or Allegro X.

And our next question comes from the line of Joe Quatrochi with Wells Fargo.

Just to follow up on another question on the China impact. I mean, I guess, can you help us understand what would have RPO been had the restrictions not been in place exiting the quarter that had been rescinded prior to exiting the quarter, just that difference, so we know what RPO, I guess, technically really is now? And then just to clarify, on the full year guide increase, is that all driven by the upside from China? Or is it other regions as well?

Yes, Joe, just to the second part of that question first. I mean the increase is because of the strength we're seeing across the board and across all geographies. I mean when we were doing -- updating our guide, the guide we gave you at the end of last quarter was without any China restrictions. And at the time we were updating the guide, we obviously knew that those restrictions have been rescinded. So it is an apples-to-apples view when you compare the guide now against this time last quarter. We've taken the year up by $50 million, and we've taken up EPS by about $0.12. And that's on the back of very strong bookings activity and performance that we're seeing right across the globe. In relation to the backlog impact of China, when we held up revenue for China, any of those orders in which revenue was paused as a result of the China restrictions as of the end of Q2, we had to back out the bookings from the backlog at that time. But I think if you look on a year-over-year basis, the right way to look at it is that we will end up the year with a higher and record level of backlog. The book-to-bill will be greater than 1 for the year, which indicates a very strong bookings half for us in the second half of this year. But that's mainly due to strength across all regions, across all geographies. And we have a high level of renewal activity that just naturally falls into Q3 and Q4 because we have a number of expiring contracts in those quarters.

And our next question comes from the line of Charles Shi with Needham.

Maybe this is for John. John, I think you reported recurring revenue as a percentage in Q2 was 78%. This is probably a multiyear low. And I wonder what's the expectation for the full year, the recurring revenue percentage? And what is the long-term normalized level? Maybe this is a related question, if I may. I believe your hardware is mostly manufactured in the U.S. and presumably, there should be no direct tariff impact, but was there any customer behavior-related pull-ins that was seen in Q2 and possibly also in Q3?

Yes, Charles, great questions. But on the hardware side of the business, I mean, the hardware demand continues to amaze us really. I mean they're tremendous products that we have there. But -- and the team is continuously trying to improve our production capability and manufacturing capability to produce those hardware systems as quickly as possible to try and keep up with that demand. We make hardware systems in North America for the North American market and outside North America for the international market. So we think our tariff exposure is quite limited. The -- yes, just generally on the strength in hardware in Q2, combined with us having to pause a lot of ratable revenue in China during Q2 caused the recurring revenue percentage to dip to about 78% for the quarter. But if you look -- typically, we look at that as a kind of a rolling annual number. We'd expect it to be about 80-20, 80% recurring and 20% upfront. And that's been growing. I mean, in the past, that was probably 70 -- what was it, 85-15 and now it's gone kind of more towards 80-20. But that's really the result of the strength in our -- in demand for our upfront businesses, which mainly come out of IP and hardware.

And our next question comes from the line of Ruben Roy with Stifel.

Anirudh, I wanted to touch back on IP. I know IP historically has been a little bit lumpy, volatile, whatever the word you want to use. But you've had quite a bit of strength recently. IP was up, I think, 30% last year, 40% last quarter, another 25% this quarter. You talked about your broadening portfolio, but it sounds like a lot of this is going into AI and HPC. And obviously, faster design cycles in those markets, et cetera, in recent years. I'm wondering if you could talk about your longer-term perspective on IP growth. Is this sort of sustainable at potentially higher rates than you've thought about historically for that segment?

Yes, that's a great question. And in general, I think I am much more optimistic in IP than I was, let's say, two, three years ago. And I mean there are multiple reasons for that. One is we are investing more in IP now because we feel, first of all, that our EDA position is very, very strong. For years, we invested in EDA, and we continue to do that. But at this point, we feel we are in a strong position in EDA. We are in a growing position in SDA, given 3D-IC and strength of Allegro and AI. And IP, historically, we didn't invest as much, but things have changed. One is because of this -- like a previous question, this emergence of chiplet-based architectures, I think, provides more opportunities for IP. Emergence of multiple advanced node foundries. There are at least four major ones now provides more opportunities for IP. And our portfolio has also improved with some good M&A, like we got HBM4 from Rambus and there are several others over the last few years. So I feel now we have crossed like a critical mass for IP to be a good business for us. And you're seeing that last year, the 1 year doesn't make a trend. I think we are seeing that this year. And so -- but I do expect in the longer term that IP can grow faster than Cadence average, which is what we like to see in the -- and it will have slightly lower margin than EDA, but of course, it can grow faster. So as a rule of 40, that's a good area that -- and we continue to invest in that. and especially with the AI-driven IPs, new foundries, this onshoring, I think it's a good business for us and good growth for the next several years, I expect.

And our next question comes from the line of Clarke Jeffries with Piper Sandler.

Just a clarification on the tax benefits. I heard $140 million for the remainder of the year. Just to clarify, is that for two quarters and that the annualized benefit might be close to double that? And then just from a philosophy perspective, does this change around R&D expensing sort of change your appetite for incremental investment or near-term windfall but normalized over time and no change to appetite?

Clarke, yes, I mean, no change at all to our approach and our strategy and our R&D investment. I mean, we love investing in R&D. We think we do that quite well. But in relation to the tax consequences of the OBBBA thing that the primary change in fiscal 2025 relates to the immediate expensing of domestic R&D. The cash tax impact of that, we get benefit of about $140 million before the end of this year. But there's a smaller portion of an impact to the GAAP P&L. Now from a non-GAAP perspective, we use an effective tax rate of 16.5%. That normalizes everything. So the impact of the OBBBA doesn't change our non-GAAP rate. for this year, it's still at 16.5%. But the onetime difference on cash taxes for the year is about $140 million. Now you'll see the benefit of that in Q4, but it's already incorporated into our annual guide.

And our next question comes from Joshua Tilton with Wolfe Research.

Congrats on a great quarter and a nice raise to the full year outlook. Most of my questions have been answered already. So maybe more of like a medium-term thought question. When you look at the guide for the full year, it still kind of implies that the recurring revenue side of the business is going to see muted growth. Now I know some of this is because there was a little bit of a hold this quarter because of China. But how do we think long term, the trajectory of recurring revenue growth from here and your confidence in the durability of total growth as maybe you roll off the hardware cycle or you start to see slower growth on the upfront side?

No. Great question. The -- yes, I think what we've seen over the last few years is we've seen a drift towards kind of a lower level of recurring revenue, higher level of upfront revenue, but that's mainly been as a result of IP and hardware and SD&A to a certain extent, growing faster than the average Cadence business. But I think -- I mean, right now, we're at 80-20. We're not guiding anything. We're at 80-20 for this year, but we continue to expect that split that we're not guiding for next year yet, but I actually think that the core EDA software is doing so well that there's quite a good chance that 80/20 remains for quite some time because we're seeing good growth there. Like we're seeing a lot of growth right across the whole portfolio of the Cadence business and across all geographies right now. So we're very, very pleased with the way that's working out. And really, the change in recurring revenue is just a slight change in how customers consume our technology and our solutions, and it's how we provide them. And we're just delighted with the continuous adoption from those customers.

And our next question comes from Nay Soe Naing with Berenberg.

And also congrats on the quarter and the raise on the full year guide. A question on Agentic AI, please. I was wondering if you could maybe share your thoughts on what do you think will be the toughest adoption barriers because from my point of view, at least the Agentic AI products that you guys have, unlike the broader software AI that we've seen, ROI shouldn't really be a sticky point here. So I was wondering what would be the sticking point here? Would it be the operational challenges, i.e., customers adopting or implementing your Agentic workflows in their established workflows today? Or is it more of a human element here where unfamiliar with the technology or people are somewhat worried that their jobs may be at risk from adopting your AI solutions?

Yes, very insightful question. I think one thing I would like to emphasize that there is a difference in chip design and system design versus general software, what I have seen versus like because this is engineering software versus kind of IT or enterprise software. And our history in EDA, I mean, there's a few things that are different. First of all, engineering software or EDA, we have already provided over years a massive level of automation. Now it was not because of AI. It was used in the past was because of classical methods. So our users and customers are already used to a lot of automation. If I look at like 20 years again to now, I think the EDA productivity -- productivity by EDA has gone up by like 100x, like things like what used to take like 500 people five years to design will now take like 50 people like one year to design or something like that. So -- or like half a year to design. So our users are already used to a lot of automation, which may not be the case in like classical kind of software. The second thing is that our workload of our customers is going up exponentially because of Moore's Law and 3D-IC. So this is a very different environment than if the workload is constant in some industry that is not evolving. In chip design, like by 2030, the chips will be -- right now, they're 100 billion to 200 billion transistors. -- is expected the chips will be 1 trillion transistors by 2030. Then you add all the software, you add the new architecture. So the workload will go up by 30, 40x in the next five years, okay? There's not even enough talent or headcount to hire to meet that requirement of 30x. So this is not an industry in which the workload is going to be fixed, okay? And then the worry of the people is if you use AI, your job will be affected. Here, you need AI to cope up with the 30x. So I believe that the customers -- and this is talking to all the big CEO customer CEOs, they will invest in R&D, okay? But they will in headcount, but they don't want to invest 30x the headcount, okay? I think the headcount in our customers' R&D will go up maybe by 2x, 3x. But the remaining gap of 10x in productivity has to be made up with more automation. And they are willing to invest in agentic AI and more compute to balance that because there's not even that many engineers you can hire. So the two things which are very different in EDA and SDA versus general software, one, our customers are used to more and more automation over the last 30 years. And second, the workload is going up so much that they have no other choice but to use automation and AI. So I think that's why the real test for us will not, in my opinion, be whether the customers are willing, I think the customers are willing, is the productivity of our solutions. So what I've seen with our customers is if the product works, if they give better PPA, if they are faster, our customers will always adopt that and because they have more and work to do. And that's what we focus on, like Cerebrus AI Studio, does it give 20% better PPA? Or does Verisium give 5, 10x benefit? And this is the history of our customers. And because these are the best and the biggest -- the brightest companies in the world, right? We all -- MAG 7 are our customers, all the top 50 companies in the world. So we deliver value, I've always seen they will adopt it because the workload that they're doing is increasing a lot.

Workload, and [ automated ] AI tools create more demand for our core EDA tools as well, right? But we always said it would take a couple of contract cycles. And we're always very mindful of the balance between delivering cutting-edge innovation and ensuring our solutions remain accessible and valuable and useful to customers of all sizes. I mean the increasing design complexity, particularly with AI and advanced nodes naturally creates demand for more sophisticated tools and IP. But our goal is always to deliver measurable ROI through productivity gains, faster time to market and the improved PPA outcomes that customers can get from using our core EDA. So we do think core EDA has growth in there. and we could well benefit from that over the next few years.

The next question comes from the line of Siti Panigrahi with Mizuho.

Great. Yes, most of my questions have been answered, but I just want to follow up to one of your comments on the strong bookings. How do you characterize the demand for your traditional semi customer versus systems like hyperscaler, that segment? And do you see any kind of positive sign on the traditional semi customer?

Yes, that's a very important question. I mean, first of all, the system companies are doing more than before, as you would expect. not just the classical data center, but hyperscalers, which are -- but also like physical AI like cars. And then on the semi side, of course, some customers are doing phenomenally well, right, like NVIDIA and somebody like Broadcom. So I think your question is more traditional semi. So I do think -- I mean, of course, this recovery in traditional semi has been projected for a long time. But I do think there is some recovery now. I mean, at least in the memory market, there seems to be in the mixed signal. And we highlighted, for example, ADI, that's doing phenomenally well, and we had a very good -- ADI is a long- term partner with of Cadence, but we had a very good expansion in Q2. So I think there are some signs that the traditional semi are also doing, but it's still early, and we'll wait and see. But for us, we are -- as you know, we are very diversified, both geographically and customer-wise. So -- it's important, but not critical for us. So we are patient. So the recovery will happen in traditional semi. Maybe some of it is starting. And to the extent it happens, we'll be ready for it. So -- but we are not critically dependent on a particular customer set recovering at a particular time.

And our last question comes from the line of Blair Abernethy with Rosenblatt.

Great quarter. Anirudh, I just want to ask you again about the system design analysis, the traditional simulation, multiphysics simulation. You're outgrowing the market pretty substantially even if we back out a couple of extra months from Data CAE, you're still mid- to high 20s, it looks like organic growth. What's driving that organic growth? Is it Millennium helping with that? And I'm just wondering, as you look at that market, which is sort of a 10% kind of growth market, how long do you think you can sustain that significant growth?

Yes. Good question. I mean, like I mentioned, the growth is driven by multiple things. I mean, 3D-IC and the strength in Allegro that pulls in other products is a key factor. It's not just Allegro and 3D-IC by itself, but all the analysis tools because I think most of the disruption in the system space is either very, very close to the chip, like with 3D-IC or it is very, very far like data center simulation. And we have great partnership and products in Cadence Reality, which is full data center simulation and then very close to the chip, which is Allegro and 3D-IC and Integrity. So those markets, I think, should be growing faster than the overall system market because that's where the disruptions are happening, and that's where we are focused on. And then BETA is helping to pull in some of the other -- because we -- one of the key channel challenges in the system side is to build out the channel. And even though not only BETA provided great products, it also helped us on the channel side, okay? So that's the second reason. And then I'm super optimistic about Millennium, and we're kind of spearheading this kind of revolution. I can talk about it for a long time. The CPU plus GPU integration with partnership with Jensen and NVIDIA and AI together. But I think it's still in the very early innings. So Millennium is still in very, very early innings. I mean we have a lot of pipeline and demand, but it still has to play out. So -- but overall, I think we'll see, but we are pleased with our positioning in systems, especially because we are positioned in the more exciting part of the system market, I believe.

Yes. Just to finish, I'd encourage you to keep focused on the kind of -- the annual kind of outlook for the company because quarter- over-quarter numbers can look a bit odd sometimes. As you called out in your question, BETA is included in our Q2 numbers this year, but wasn't there last year.

And I would now like to turn the call back over to Anirudh Devgan for closing remarks.

Thank you all for joining us this afternoon. It's an exciting time for Cadence with strong business momentum and growing opportunities with semiconductor and system customers. With a world-class employee base, we continue delivering to our innovation road map and working hard to 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.

And ladies and gentlemen, thank you for participating in today's Cadence Second Quarter 2025 Earnings Conference Call. This concludes today's call, and you may now disconnect.