Keysight Technologies, Inc. (KEYS) Earnings Call Transcript & Summary

We'll start it off. Maybe, Neil, I'll start with you. Maybe just give us a little -- it's been a pretty good run of recent. So maybe just high-level kind of what were some of the positive factors in '25 and what's informing the pretty good outlook you gave for next year?

Yes. That's a great question. As I look back on '25, the first thing that comes to mind is building momentum, right? We had started the recovery as we were entering '25, but we really were talking about a mixed demand environment. Some of our businesses up, some of them still a bit soft. And we've really built momentum as we move throughout the year. We started with growth expectations for '25 that were at the low end of our long-term range, 5%. We took it to 6% after half, 7% after 3 quarters. We finished at 8%. And then we've guided to 10% growth here in Q1. So we really have built momentum as this recovery has happened and as our end markets strengthened. And so I think as we enter '26, we had a great order book in Q4. We have a strong funnel, strong backlog going into the quarter. And maybe more importantly, we've got these big technological themes that are aligning positively for Keysight. I think number one among those is the impact that AI is having on our wireline business in commercial communications. But in addition to that, our wireless business returned to growth in '25, and we see that growth sustained into '26, driven by things like nonterrestrial networks, early 6G research, Open RAN, these types of things. Semi is rebounding, some of that driven by AI, but also looking for assurance of supply in the semi markets. And then, of course, aerospace, defense, we're also bullish there where not only are U.S. defense budgets up and talking about things like the Golden Dome and those types of things, but you've got your European allies committing to spend on defense at a higher percentage of GDP. So we are -- we've given a soft guide for '26 that we expect the growth rates to be at or above the high end of our long-term model. We expect to deliver 10% or more EPS growth. And so we're well positioned here as we enter '26.

Okay. Great. Yes, it's a good starting part. We'll dig down on a lot of the topics you talked about. Kailash, maybe we'll get you involved early here. Last quarter, you did give us a little more color on kind of AI influence on the wireline business, about half of it related to AI. So -- maybe if you could start off by just walking through kind of the key use cases that you guys are addressing in the AI world. And I'd love your sense. We know the wireline business has been outgrowing wireless. So if you could just touch on like the impact that AI specifically is having on the growth in that business?

Yes. We're -- we've talked about accelerating technology megatrends. And clearly, AI over the last 2.5 years has been a big growth driver for all of Keysight, definitely commercial comms business. So as the industry has sort of gone from experimenting with AI and deploying this at scale, you think about every design that goes in compute, networking, memory, interconnect and power, all of the products, all of the systems that need to be designed to enable those dimensions, get tested, emulated and designed with Keysight solutions. You talk about high-speed switches. You talk about interconnects, chip interfaces, custom silicon, DSP. Majority of these products are being enabled by Keysight solutions. So we're helping lead customers from the chipset folks to help them test these interfaces at high speeds. We enable component makers that get deployed in data centers in these scale-up and scale-out networks with our capabilities. We're able to emulate optical signals, electrical signals that they would be -- that those components would be faced with in a data center environment. We enable them to test fidelity, power, these types of parameters. Then you kind of go up the stack, you have these switch makers. You think about the Aristas of the World, you think about the Juniper networks of the World, we're enabling them with emulating AI traffic as they design their products in the lab environment because that's what they're going to be seeing -- their products are going to be seeing in the AI data center infrastructure. So we help them there. We have contract manufacturers and others who are building these accelerator cards, AI offload accelerator cards, GPU cards, and we're helping with a lot of our basic products to help in that production and the manufacturing flow. And finally, we have hyperscalers, and they're doing a lot of different things, obviously, at the top, they're emulating workload traffic. They're training models. When these models get trained, you have workloads that move across AI racks, servers and clusters. They move at different speeds, 400G, 800G, 1.6 tera and beyond. They're also looking for how well their GPUs are utilized; how much time it takes to train a model. What is the benchmarking? How does it compare using this type of a rack or that type of a component, we're helping hyperscalers in their ecosystem with the new solutions that can actually emulate workloads as these models get trained. So those are some new capabilities that we introduced in beta form last year. And in this year, we made formal shipments of those -- some of those products, the Keysight AI data center builder in particular, has received a lot of industry awards. It's one of a kind, and we're leading the industry with that capability and helping hyperscalers and their ecosystems with some of those needs. So if you pull back up, I mean, we've got the broadest stack all the way from physical layer solutions to these application layer solutions, allowing benchmarks of -- benchmarking of AI workloads, and we're able to cover everybody from chipset players to component makers to subsystem makers like the switch makers, the ODM, CMs and the hyperscalers. And that's also leading to a widening of the entire customer base as more and more companies, just the traditional players, but also a lot of start-up companies entering the fray.

Right, right. Yes, it's just a lot of different vectors you guys are participating there. So just curious what you kind of see moving forward, like if we sat here 3, 4 years ago, I probably barely knew what a back-end network was. Now it's huge. Now we're talking a lot of scale up. We're talking scale across. So are you still at the phase where you think a lot of your customer base will be still expanding with new companies and new use cases that are going to develop over the next few years to further supplement the growth here?

Yes. You think about the gigawatts that are getting deployed, right? There's another 10, 15 gigawatts that are going to get deployed potentially in '26. That's the projection and then another 100 gigawatts of capacity over the next 5 years. Scale up, these are things that happen inside a rack. These racks are going from 10 kilowatts to 100 kilowatts. There's more complex networking, switching, routing that goes on. And then there are protocols that define how the workload needs to move across the different GPUs and servers. So that's an expansion. Some of that is not quite standardized, but there are new standards coming in, which means that for the scale and the standardization, you need to have interoperability type of capabilities. So you can say, okay, we deploy this component and then that switch, things still work in a rack. So those are additional opportunities for us in the scale-up space. But then there's also the scale, right? I mean you've got to deploy all of these racks. So we participate in that, too. You go to scale out, that's all about networking and connecting these racks inside an AI cluster. Some of these clusters are hundreds of thousands of GPUs and other compute capabilities. Again, there are new protocols like Ultra Ethernet Consortium Alliance is coming up with on scale-out networks. You have transceivers that need to get built at different speeds. So that's an opportunity. Scale across, again, as data centers get disaggregated, distributed. And as the data centers move closer to the edge, that's an additional opportunity. So we're pretty excited about all of these vectors. But as with anything that's sort of ramping this way, there are periods of capacity digestion that's likely to happen, but I do believe that this is a secular trend.

Okay. Yes. Maybe the last one on this topic. You mentioned that maybe, Neil, you can jump in the emulation and simulation software type of solutions seems to be growing really well. You've obviously made some acquisitions and built some of it yourself. Maybe, Neil, talk about the importance of maybe transforming the model. I think you're close to 40% now software and services. Just kind of update us on how you see the move to software and some of these newer services impacting the model going forward.

Yes. I mean I think this has been a multiyear journey for Keysight, really starting from around the time of our spin, maybe even before increasing the value delivery via software. We made some great inroads and then there was a big push to move towards more recurring models. And I think we still have work to do there, but we've made a lot of progress. Our business is about 30% recurring revenue today, as you noticed, about high 30s percent software and services. We do believe that the acquisitions that we just completed add about 3 points to that. So we should be in that 40% range here on a go-forward basis. And then I think as Kailash has mentioned, we continue to find new ways to deliver value through software. And so I think organically, that mix shift tends to be relatively slow because our other businesses are growing as well, right? So -- but then there are opportunities for us to accelerate that mix shift via acquisitions, not just the ones that we just did a couple of years ago was ESI. That was a pure software business. The years before that, it was Eggplant. And so there has tended to be a software bias towards our -- to our M&A efforts, which is helping us to accelerate that shift. And you see it in our gross margins now in the mid-60s, whereas 10 years ago, we were in the high 50s. And so certainly contributing to the overall profitability of the business as well as the stability with the recurring revenue components.

Okay. Great. Maybe if we transition to wireless, obviously, that was a big driver for Keysight several years ago when we had the massive 5G deployments and R&D phase. You mentioned nonterrestrial networks and O-RAN and early 6G. But could you just kind of frame that how should we think of the time line of the wireless business? Because at some point, we're going to really be inflecting on 6G, I would think...

Yes.

Like we've seen in prior cycles. It's obviously been a little bit more delayed than prior cycles. I think maybe 5G didn't do what the operators needed. So it's a little slower. But walk us through kind of how you see the next few years and when we could see more of an inflection in that business?

So you look at 5G CapEx peaked in '22. Obviously, there's been a little bit of a reset in the industry. Now all of the companies that's part of the wireless ecosystem have started to post better financials. There's smartphone subscriptions have gone up for the first time. It started late last year, and it's continuing. There's some excitement about how AI models can drive new devices, but also existing devices like the smartphones with better applications and so forth. So there's some excitement building. And the better financials are just allowing our baseline of customers to invest in ongoing 5G standards, right? So Release 18 and 19 are underway. I mean that's all called the 5G advanced umbrella. So triggered mobility, which drives power efficiencies, enhanced uplink where now these devices are generating a lot of data. It needs to get into the edge or the cloud. So some of these capabilities have to be installed, which means that there is ongoing R&D with these standards. We also see ongoing fixed wireless access networks deployment on the network side. Clearly, Verizon has been doing that. Now you also saw T-Mobile talk about that. So there is ongoing activity from a network rollout perspective. So all of this is sort of leading to a recovery in the base business. And when subscriptions go up and stuff, you have these component makers also benefit and they have to design new components, whether it's for power efficiency or getting deployed in a newer band or a newer form factor, they have to use our capabilities, both in the design workflow as well as deployment validation. So there's that aspect. The second aspect is NTN. NTN nonterrestrial networks, it started out as a direct-to-sell sort of service for emergency purposes, but now it's starting to become a mainstream application or there's work that's going on to make it a mainstream application. The number of LEO constellations are growing. And if you kind of think about that, each satellite costs north of $700,000 and sometimes over $1 million. So the amount of design, emulation and test that needs to go into building that capability is driving R&D investment as well as manufacturing deployment investment from our customers. We have a very complete portfolio [indiscernible] we're able to -- and satellites are not just satellites, you have to ensure that it's communicating with the ground infrastructure. So there's investment going on in ground infrastructure. There's investment going on in customer handsets. There's investment going on in that communication over the channel and the constellations themselves with -- and Keysight has the broadest and deepest portfolio to enable that application. And now with the acquisition of Spirent, we get satellite constellation emulators, and we're excited about augmenting our current portfolio with that. So this is an application that will grow over the next several years. And then 6G, we started to see an inflection. It's small dollars, but it's moving from pure research to prestandards development and activity. Verizon announced an innovation forum, a 6G innovation forum with all of the traditional wireless players like Nokia, Ericsson and Qualcomm, Samsung, but there are also new players like Meta that are being part of that consortium, and there's some activity going on towards that LA 2028 event. There was -- in Prague, the 3GPP standards meeting was held. They talked about 10 different focus areas for 6G. So all of this is driving some prestandard activities. And what we do is we allow customers to test and build and design these capabilities to see what works, what is going to work during commercialization. We help them test that in advance in their labs, so that they can shape the standards. So all of this is going to drive some activity in 6G. That's still towards the end of the decade, but we expect a steady ramp towards that. So those are sort of the vectors, the traditional device space with AI in networks and AI in devices sort of driving activity, ongoing standards, nonterrestrial networks and then these prestandard 6G activities kind of driving our wireless business.

Okay. Great. Maybe if we shift to kind of the defense side, defense automation. Neil, you touched on it in the beginning. Maybe just start with like kind of what have been the impacts of whether it's a shutdown or DOGE, like have you guys been impacted by that? And then on the go-forward basis, you have not just a heightened focus by the U.S., but a lot of the European nations as well investing more in defense. So maybe walk us through kind of your view of how that business is positioned over the next few years?

Yes. Maybe I'll start and then I'll let Kailash comment a little bit more. I mean I think we look back at '25 and say this -- our aerospace defense business put up record orders in the face of a number of challenges, right? So first of all, any time you have a presidential administration change, you tend to see pauses in those markets. And then you had DOGE. We would characterize the DOGE impact as relatively small, but not 0. And you had continuing resolution and most recently, this government shutdown, which, again, not huge impacts, but not 0 impacts either. And so I think the fact that we went through '25 in the face of all that and still deliver record orders is a very positive sign. I think as we look forward, the trends are pretty clear. I think there tends to be broad political alignment in the U.S. on the need to invest in defense technology. We've seen that now reflected in defense budgets going back to the start of the first Trump administration. So that's across presidential administrations of both political parties. And then more recently, you've seen our European NATO allies commit to spend on defense at a higher percentage of GDP. So I think all of those present potential tailwinds for us as we look forward, and I'll let Kailash potentially provide some more specifics.

Yes. The defense modernization trends and strengthening deterrence trends, it's just -- it's ramping. We have clearly the traditional primes that are investing in Europe and U.S. But there are new programs like Golden Dome, like the GCAP program, which is a sixth-generation fighter program. There are areas in space where LEO is used, and coexistence of LEO and GEO satellites have to be thought about. There are those new applications, drones and radar, these types of applications are driving business. We saw strong growth in Europe. And when you kind of think about what capabilities we offer, I mean, it spans a wide variety of use cases, right? We allow -- we provide capabilities to test phased array antenna arrays, help with the design and test of that. That goes into satellites, that goes into many ground type of capabilities. We have solutions that emulate radar target generators. These are new capabilities that we've been able to build, combining many capabilities that we've already had for new use cases, and this is helping us win new awards. We're able to provide capabilities for tactical communications. These are handset type of communications. We're able to help customers design and test those. We have -- we're also moving up the stack. It was an event, Cyber Quest '25, where we provided -- we tested the resilience of the U.S. Army communication infrastructure by emulating millions of concurrent users to see if the network would be resilient to security vulnerabilities and things like that. So we're able to integrate some of our Ixia acquired capabilities, some of the capabilities that we use in AI data center infrastructure, combined with our physical layer portfolio to enable solving some of these problems. Cybersecurity, threat emulation, multiple use cases here that are addressing many of these use cases, we're seeing defense technology start-ups also join the fray and there are about 1,400-plus defense technology start-ups around the world. And they're involved in software-only digital twin type of modeling of command and control, but then there are also companies that are making hardware gear that need our capabilities. So that's actually another driver. So Europe is a driver. I mean, if you think about U.S., it has always been a driver over $1 trillion of defense budget next year. The RDT&E line item, which is the R&D test equipment line item is growing 20-plus percent. So that's there. And then you have Europe. And then you have these new defense technology start-ups and new programs coming online. So it's always a business that's hard to call on a quarter-by-quarter basis, but it does present a secular tailwind for us as we go into the next.

Okay. Great. I do want to follow-up on the semi comments before. Maybe just talk a little bit about that business. I think there's obviously a different set of competitors that you may come across in the semi test side. Talk a little bit about the trends you're seeing there and what is Keysight's differentiation on the semi side?

Yes. So if you -- when you talk about our wafer test capabilities, so we test the capacitive dielectric properties of every -- is the wafer really suitable for making a chip is what we do, right? And we're able to deal with multiple -- think about things over 25-nanometer all the way down to 3, 2-nanometer and each is a new SKU, new update that our customers need to procure. And clearly, with an inflection in AI, we're able to capitalize on that. Silicon photonics is another area where more and more optics are getting plugged into a wafer level capability. So that's a new set of challenges that our customers have. You have to test and design for electrical to optical transitions. Are you having any signal loss? All of this needs to be evaluated way upfront before a chip can be made. So we're seeing activity there. We're also -- we also have software. I mean, you're talking about semi, but we have -- all of these have to be modeled from a software point of view, and we're -- we've launched capabilities to model all of this photonic integrated circuits, how do you model this from a design perspective? So we've got that, and now we're getting some capabilities from the Synopsys divestiture of Optical Solutions Group. So we've got that capability. We also launched a wireless Pro solution, again, to model and test things at the silicon level. So we have our classic wafer test solutions, but we've got augmentation with our software simulation tools as well that's helping from an overall semi point of view.

Great. Neil, maybe we'll go back to you with more of a financial question for you. I think you mentioned earlier the good growth rate outlook and this 40% incremental op margin. I know there's moving parts with the M&A this year. But just walk us through kind of how you view operating margin potential as the acquisitions are integrated. And if there's more success on the top line, how should we think about ability to pull that to the bottom line?

Yes, absolutely. So I appreciate the way you answered the question and kind of distinguishing between our core business and what we have going on in the M&A space. I think both present exciting opportunities for us. I think as we think about our core business, we put out a model, which we've managed to over time, which is in our business is growing mid-single digits or better. We look to deliver 40% operating leverage essentially. And so obviously, to the extent that growth is outpaced as a result of either a snapback or accelerated growth in AI, there's potentially opportunities to outperform that 40% incremental leverage. On top of that, we just completed 3 acquisitions in the month of October, and we're working very hard to get those integrated. As we inherit them, they're operating at operating margin levels that are significantly lower than Keysight's high 20% operating margin level, but we expect over the next 12 to 18 months, we're going to get them to the point where they're going to be accretive to Keysight operating margin. And so we put out a cost efficiency target of $100 million. The vast majority of that is synergies from these acquisitions, although there is a portion of that, that comes from other efficiencies that we're looking to drive across the Keysight portion of the portfolio. And again, I think that's going to offer us an opportunity to drive exciting leverage over the next couple of years. We did say that we expect these acquisitions to be mildly dilutive here in '26, but accretive to earnings in '27. And even with that dilution, we still expect to deliver EPS growth that's at or above our 10% target here in '26. So we'll overcome it based on the strength of our core business and some of the market opportunities that we've already talked about here today.

Okay. We've got a minute. So I think we'll wrap it there. I appreciate the overview here. I'm always taken it back by how many different technologies and use cases you guys are involved in. It's pretty amazing how broad the technology is at the company. So it's nice to see it being rewarded lately. But thank you.

Yes. Thank you, Tim. I think -- we're excited about the future and the opportunity ahead of us over the next couple of years.

Excellent.

Thank you.

Thank you, everybody.