Kyndryl Holdings, Inc. ($KD)

All right. We're going to get started. Thanks, everyone, for joining. My name is Tien-Tsin Huang. I'm the IT services analyst here at JPMorgan. And really happy and grateful to have Martin Schroeter here, CEO at Kyndryl, to join us and have a fireside chat. I've taken a lot of questions from the investment community, Martin, and we'll go through them over the next 30 minutes or so. But thank you for being here. It means a lot to me.

Thank you, Tien-Tsin. Delighted to be here. I appreciate it.

I know how busy you are and a lot of demands on your time. So I'll be efficient, but thinking about how to start the conversation, Martin, I know you're always on the road, you're meeting with clients. You're talking to CEOs, CIOs, boards, what have you. You guys touch a lot of IT estate across large enterprises. So what are you hearing from those counterparts that I mentioned? What's been changing? How are you changing the strategy to address what you're hearing on the ground?

Yes. It's a great question. Good afternoon, everybody, and thank you for joining us here in person, Brendan, nice to see you. I guess there are some, what I'll call kind of long arc themes that evident in nearly every customer conversation. A little bit different. And as you said well, we run -- we run a lot of workload -- regulated workloads, right? We're in mission-critical. So the way we feel things is a little bit different than some others. But let me go through sort of what I see as kind of consistent themes with our customer base. First and foremost, both cybersecurity and resiliency remain top of mind. And even today with AI starting to have a bigger and bigger impact on how the world works, that is not going to change the focus on cybersecurity and resiliency other than to say it's going to go up another notch again, right? So the idea of securing yourself, but more importantly, the idea that you can be resilient, which regulators have now taken up in different parts of the world, European banking, for instance. So cybersecurity resiliency, a long arc theme, where we play already and is going to continue. The second I would put into the sort of the context of all of the new innovation that's coming out, we'll just throw AI as the sort of one of those things. But AI and and all of the technologies around it, they really lead to a discussion around modernization. AI -- one way to think about AI metaphorically is that there is just now the shiny new -- the shiny new bullet train that can go 200 miles an hour, but most companies still have tracks that were built for the 30-mile an hour era and to modernize to get ready to use that AI takes a lot of work. And so our customer base is keen, keen, keen to figure out how to modernize, recognizing that these are not systems that you can take offline and modernize them and then bring them back up, right? This is very much a run transform run world, which is what we do very, very well. So security resiliency still top of mind. Modernization is still top of mind. And then we get into what is ever increasingly difficult, there's the sort of the geopolitics. And the reason it's important, particularly in what we do is for instance, in many parts of the world, the idea of sovereignty is top of mind. You cannot have a discussion with a company in Europe, most parts of Asia, South America, Latin America, without talking about sovereignty. And sometimes it's data sovereignty, sometimes it's AI sovereignty. So the idea of sovereignty is making the world more complex again. And at the end of the day, sovereignty for our customer base is really about control. And how do I make sure I understand my -- the status of my systems today? And how do I make sure that I'm building something that will survive that -- whatever the geopolitics are of the day. So for us, again, it translates to a more complex IT estate, a more hybridized estate. And we're seeing, in fact, an increased interest in private clouds because of -- just because of the sovereignty issue alone. So we have all of these sort of, I'll call them, again, long arc trends. They're not going to change. Cybersecurity and resiliency is going to be important for a long, long time, forever, I mean, we could probably say forever. We know that that there are always going to be systems that need to be modernized. Yes, today, it's driven by AI. It's driven by new data technologies. And we don't know the geopolitics. But again, given the role we play in the world when our -- our customers are looking to modernize, they also want to be able to modernize for something that's going to give them comfort for the next 4 or 5, 6 years and adaptability that they need. So I would put those 3 as the big sort of themes. In any given region, it could be a little bit different. It may be weighted a little bit differently, but those are the big 3 themes that you cannot meet with the customer without being ready to talk about.

Good. No, that's a good summary. But I think on the call, I just jotted it down as you were calling it out, Martin, just you described some of these extended decision cycles due to some of these same things, sovereignty. I think you also talked about AI choices and some of the regulatory complexity, which you could argue maybe as part of the whole geopolitical piece. So we talked about this a little bit before we started, right, Martin, how much of this is just the new normal? And I know we've been waiting for inflections and pivots and changes in discretionary spend. But how do you see the change that could trigger up or down [indiscernible] spend?

Yes. I think that the things that I identified on both on the call and again, I'll put sovereignty back on the table, this is now the new reality because every company operates within a legal framework, driven by the country that they're in. That country is responsible to a citizenry, et cetera, you know this as well as anybody. So again, given the role we play in what we do, I think this is the new norm of decision-making. How do I make sure we sign contracts 5, 6, 7 years? So how do I make sure that I have the flexibility I need with the partner who I really need, who can lead me into that new world. So I would say, to use your shorthand, I would say that is the new normal. How do I really think about where I want to be for the next 4, 5 or 6 years. There are always in the minds of our customer base, all customers, all enterprise IT runs on a business case and a lot of those business cases are driven either by productivity or growth. And so that tends to be more fed by macro. So is that the new normal? Well, this macro will change again. But right now, people, I think, are feeling like I really need to be in AI or really need to figure out for my company, which means I need to make investments in modernization but that can also change again in the future, right? As the promise of AI starts to be realized in more and more places, I think we'll see that shift. As we sit here today, 99% -- I'd like to say all, but I won't use the superlative. Nearly every business case that we help our customers build on AI is productivity driven, productivity driven. And to give you an example, we we agentified, I'm not sure if that's a verb yet or not, but we introduced agentic AI into one of our banking customers, KYC, know your customer onboarding process. And we will save them 30%, 40% will make it a lot faster for them, but that's how they did their business case. They also believe very strongly, though that they were able -- because they were able to take the time in their KYC process to onboard a customer from, call it, 25 days to about 25 minutes, they believe that, yes, they've built their business case on productivity, but that will lead to more growth. And I think as we prove to our customers, as our customers prove to themselves that it's not just about productivity, that there will be more and more of a growth component coming to AI. But AI, again, has to be that technology works, but you have to get your employees ready for it as well. You have to get new ways of working. So again, I go through all of that because while there are things that are part of the new normal sovereignty as an example. New technologies are always going to be coming in. There are some that are driven by macro. And we'll have to see how AI starts to prove itself in real use cases as it makes the shift from being, again, nearly entirely productivity-driven to being growth oriented as well.

Okay. Yes. Well, the models are improving. You gave a good example. You also did mention that April was a good sales month with a couple of deals closing. And so that was promising. I don't know what -- was there a moment maybe where things got more clarity and you saw activity happen in April? I'm just curious how you -- we should interpret that?

Yes, look, I think -- I will tell you how I interpret it. So a couple of things. We did see a number of deals move a little to the right as we came out of the last quarter out of the end of fiscal year. Not a surprise, again, because of the nature of what we do, because of the many, many, many constituencies involved, regulators, employees, management, CIO, et cetera, because you have to bring all these together at a certain time, it's never a surprise to us if a decision gets moved by -- sometimes it's by hours, sometimes it's by weeks. And in April, we happened to close a bunch of deals, and we actually got more again closed that will close for the year. So we did have a very good start to the signings year. As you and I have talked over the years, we've always been focused after we went through our focus account process, and we were very focused on engineering a decline and getting certain content out. we've been focused now on making sure that the signings number exceeds revenue over time so that we can get back to a consistent level of growth. And again, I feel pretty good about where we are now and what we've already gotten done in the quarter. And I look at our pipeline and our pipeline is already better than it was last year at this time. It's a pretty good mix of new content in existing customers and entirely new customers as well. So a good start to what is -- every year is an important year, but a good start to a very important year for us.

Good. And I know you and Lori and team and Harsh have always said, right, don't spend too much time focusing on signings. And of course, the quality of the signings do matter. Now that you've been through some of the cycle and you mentioned the pipeline is good, is it safe to say the quality is what you're looking for, whether it's book-to-bill or gross profit book-to-bill short term, long term?

So a few things. So the short answer is yes. I'm pleased with where the pipeline sits, and I'm pleased with it for a number of reasons. We have, I think, been very focused, and you know this, we've been very focused on improving the fundamental profitability of Kyndryl, and I think we've made a lot of progress. Additionally, I think the agentic world and our own experience with agentic suggests that we have really 2 things going for us within our business model. One is that because we're really kind of an outcome-based services business, right? We have to deliver uptime. We have to deliver resiliency features. We have to deliver security features. We've proven over the last 4.5 years now as we automate things that, yes, we have to deliver some savings to our customers. They're looking for productivity, but we also get to keep some of that ourselves. And what I think is exciting for us in this more outcome-based business model is that agentic can help that quite a bit. We have, as we sit here now, 1,370 agents in the infrastructure. This is not the agents that our staff build to help them perform their jobs. These are agents in the infrastructure, making us more productive. making us able to automate things, allowing us to deliver higher quality services. And again, as we introduce more and more agents, yes, we're going to deliver a value prop to our customers that's really meaningful. I mentioned the bank example, but we also have proven that we get to keep that as well, some of that as well. So when I look at the pipeline, I'm excited about the content in it because it is becoming increasingly making more and more use of our own agents. It's making more and more use of our own automation, which, again, we've proven we get to keep a piece of. And then it has a bit of a mix to it that I'm really pleased with. One is, as we've invested in more and more capabilities, we've been able to expand the scope with our existing customers. So as you said well, a few minutes ago, certain signings mean a lot to growth and certain signings could just mean you've renewed something and that's not going to change the trajectory. It's good you got it renewed. You didn't lose it, but you're not changing the trajectory. It's always been important for us as we've reoriented to growth to focus on expanding those relationships and expanding the future revenue growth opportunities for us. In a place like the U.S. where sovereignty is not at all an issue, right? So we never slow a discussion down about sovereignty in the U.S. We got the U.S. back to growth already last quarter, and they've got a good growth profile ahead of them. So some of that's driven by new content within our existing customer base and some of it is driven by new customers who -- they are now more and more increasingly comfortable with us as a services provider. Many of them would have either had a little bit of experience with us in -- with one of our hyperscaler partners or more likely, they've had a little bit of experience with us in the consult side, but now that can turn into a much more robust, bigger, more substantial managed relationship. So pipeline is bigger. It's got a good mix. It's becoming increasingly -- it's going to increasingly take use of our own agentic, which allows us not only to deliver a great value prop, but we get to keep some of that as well.

So is that the answer to the phasing of the year when we think about first quarter being the trough and second half having some acceleration? Is that the answer to the phasing?

Yes, it's some of it. Some of it is certainly driven by the pipeline. We've also made a decision to take a bunch of cost out of our business. So from a profit perspective, the first quarter absolutely will be the trough, and that will then start to pay back already second quarter and obviously, second half and then give us a big lift into next year. And from a revenue standpoint, the short answer is yes. Our timing of looking at deals when they were signed and what our pipeline looks like says second half will be the stronger half relative -- well, within the year, first half, a little bit slower, second half, a bit stronger.

Good. Okay. So we'll get into some more of the details. But I think, when people are submitting questions for me to ask, the IBM piece did come up quite a bit. I know that's a part of the story and the visibility. So just remind us there with the commercial dynamics. I know that's putting pressure on revenue in signings, but it's not impacting profits. That's very, very clear to us. But just tell us about the visibility. And I know you've called out a 3-point headwind in the charts. How do you see that -- those lines behaving, let's say, in the next 2 to 3 years?

Yes. So we did all your data -- you always get your data, right?

[indiscernible].

So we did expect that by now, if you had asked us a few years ago, what's the long-term impact and again, focus accounts, our goal was to remove a lot of the hardware and software content from IBM and have them go direct and again, hugely successful. It's a part of what's allowed us to improve the profitability. And we would have expected now that customers would make sort of a consistent set of decisions, which would -- basically we would have thought it would have eliminated the headwind, but our experience last year was that now it's much more complex than just that. And it doesn't take a lot of -- it doesn't take a big number of customers, 2 or 3 customers say, instead of getting my mainframe and my software through you, even though I understand why it might be more appealing, I'm just going to buy direct. I want a relationship with the owner of the IP. Again, for -- as you said, well, yes, it's a signings and a revenue headwind, but no impact to profit because we have no ability to mark up their mainframes. We have no ability to mark up their software. So we would have thought it would be behind us. It's not behind us as we enter this year given the choices that customers made last year. We see another 3-point headwind. Look, over time, it has to start to diminish because when we were first spun out, our hardware and software built to IBM was about $4 billion. And last year, it was under $2 billion, just under $2 billion, right? So we've taken a lot out. There's probably a few more customers, I would assume this year, who were going to say make a similar decision, I'm going to go buy my hardware software directly from IBM. Again, we're completely cool with it. But it will, over time, diminish. But it does, as you said, well, it just -- it makes the story a little bit more complex, I think, than what some people would like, which is, well, just what's the headline growth rate and what's the growth rate. And for us, we printed last year down 3% and the IBM piece was about a 3.5 point headwind. So one of the things that really matter to us, we had some growth for the full year. And this year, we guided to down 2% to flat and with, again, a similar 3-point headwind. So we're kind of guiding to up 1% to up 3%, kind of, again, outside the content, but we'll have to see again how customers start to make that choice. So other than us estimating the IBM impact and the length of time, other than that, I'd say that the rest of the business is kind of performing the way we would have thought it was already a number of years ago. So we're kind of on track with where we thought we would be.

Yes. Okay. No, it's a good lesson. We'll focus on it more ex the IBMs. And like I said, the profit story is different than the revenue in the signing story. Okay. No, thanks for going through that. So let's -- time is going quickly. Consult. Let's talk about consult. It was a double-digit grower for you in fiscal '26. You exited at a lesser rate. What's assumed in the the '27 outlook? Is it hard to replenish growth in consult?

The way we built our guidance, let me start there. We do need consult to continue to grow, but we can actually deliver the revenue, the minus 2% to flat, again with the IBM headwind. We can do that if consult is growing sort of low double digit to even high single digit. We still make it, right? So we didn't want to overly rely. Yes, we're, of course, if you ask the consulting, what are their growth targets, they're far in excess of anything I've just imagined, but we also had some uncertainty around all the things we talked about, what does macro really look like and what does the sovereignty discussion really look like? So we don't need it to grow at the same rate it has been growing. And look, it's much bigger, obviously, than it was when we started. So the dollar amount that it lends a similar dollar amount even though the growth rate starts to slow. But we see and we hear from our customers an incredible demand profile around, again, the things we do, the idea of modernization, given the challenges our customers have is very real, very top of mind, very front and center for them. And in each of the examples I've used like the banking example and agentifing, and we have a similar -- for example, in government where we've done this with the government of UAE, we just announced a deal with a state government here where we're going to modify their DMV system. Each of those has a consult component to it and then a run component to it. So consult is still a very important growth vector for us. And we're -- our guidance is not built on it growing like 20%, 30%, 40%, 50%. It really just has to continue to perform on a consistent basis, and we see the pipeline for it. So we're pretty comfortable right now, we're pretty comfortable with what we see in order to deliver the guidance.

Good. And then on the managed services front and the underlying assumption there, should we assume any kind of lockstep?

So the managed services component, we see, I'd say, 2 things. One, it is where the IBM impact sits, right? So -- and since it's 75% roughly of the business, it has outsized impact on it, right? But outside of that, we've said now for 4.5 years since we [indiscernible], we said that over time, because of the shape of the backlog we inherited, we're going to go have to rebuild that backlog. And we're starting to get to the point now where Managed Services is starting -- it's stable, right, not back -- we don't need to get back to growth yet, but it's stabilizing. And over time, it is a growth opportunity. I think the work we do around managing more than half the world's outsourced mainframes and managed mainframes, that's ultimately going to be a growth vector for us, low single-digit growth, but it can be a growth vector for us. So I see the managed business is stabilizing and starting to come back. And over time, this will be a growth part of the business as well.

Okay. Good. So hyperscalers, I think, has been a big driver. Obviously, you gave a lot of great statistics around that. But the penetration rate is higher, Martin. So I'm trying to think about what's the next leg of growth on the hyperscaler side -- are you doing something differently to spur growth or to amplify growth?

There is still a lot for us to get done here, right? So again, if I step back and I look at the sort of the big picture, IBM was $4 billion of our spend down under $2 billion now. Hyperscalers were basically 0 when we were spun out because we didn't have relationships with them. And very quickly, I think the team did a phenomenal job of building what is essentially a nearly $2 billion business, right? So we sort of cross that. It's not a vitally important sort of statistic, they're not related necessarily except by what it tells us about what our customers think about us, which is, I believe, we've repositioned the business now to very much be part of our customers' future as opposed to just part of their past. And so we see future growth with hyperscalers really coming in 2 ways. One is we have a lot more customers that we have to pay penetrate still here. And most -- every one of our customers has picked a hyperscaler, sometimes 2, many of them have commitments that they need help with that to consume that over time. So we still have a lot more to penetrate within our customer base. And then importantly, because we have -- just so it's clear, we're not taking the consumption of the hyperscalers content and moving that. That's not what we're counting as hyperscaler business. We're talking about the services we've built around the hyperscalers that our customers need in order to run workloads on a hyperscale or cloud. So we have security services and resiliency services and data services that's really what makes up that nearly $2 billion of revenue. And because those are the platforms on which they're growing, those are growing workloads for us as well. So not only are we sort of still underrepresented across our customer base for that content, but they are attached to the growth vectors that our customers are on. That's where their incremental dollar is going. And that's why the message that we've been clear about is that, that's why we feel like we've repositioned to very much be about their future, not just about their past.

I like that phrasing, it's helpful to think about it because -- yes, let's come back to it. Thinking about hyperscaler, more room for penetration, you also talked about private cloud. And the sovereignty piece makes a lot of sense of why you're seeing more private cloud. But what are you advising your clients there with respect to private cloud, hyperscalers, obviously, the consequences in terms of how you staff, but what is that dynamic look for you?

So what we're seeing in our customer base is an increased interest in private cloud. Sovereignty is part of the driver. Some of it is around just wanting to make sure that they can answer a regulator's question, as you move certain workloads onto a cloud, the key -- the key question from any regulator is going to be a little reductive, but the questions that regulators ask are, where is your data right now, like exactly where is your data right now and who can see it, who has access to your data. And as workloads have moved particularly systems of engagement workloads have moved on to public clouds, that's a little bit easier to get your head around. It's a little bit easier to answer those questions. And some of those workloads aren't regulated like, let's say, a banking system or an insurance business. So what private cloud allows you to do is not just take advantage of the innovation cycle and delivery of cloud but it also allows you to put more regulated kinds of workloads because you have the control and you can answer the questions. And then on top of all of that, you can be more confident that in 3 years' time or 4 years' time or 5 years' time that whatever the geopolitical environment is, your private cloud will probably be okay. So it's multifaceted, but we do see an increased interest again in private cloud because it's around control. And you still get many, many, many of the innovation delivery features that you would expect out of a public cloud. And keep in mind, from a cloud perspective, the public clouds are delivering a lot of innovation. AI is a good example. But for customers to really consume it, it takes a lot of work, right? So you probably can't keep up at too fast a pace with all the innovation that's being delivered. And therefore, a private cloud can suit your cycle time for innovation as well. So yes, increased interest in private cloud.

Yes. So from an economic standpoint, implications to Kyndryl?

Look, for us, anything -- complexity is our friend here, right? We really get paid -- we get paid for 2 things. We get paid because our customers trust us, and we get paid because we can help them manage their complexity. And so to the extent that that any of our customers decide that for certain workloads, we can see why we want to put them on to a public cloud, take advantage of that. And for other workloads, we want to build a new private cloud, take advantage of that, anything that adds that complexity says they need Kyndryl more. So all of these things tend to be a tailwind for us.

Okay. Good. Let's -- less than 7 minutes left. So let's do workforce rebouncing then it -- just to make sure we hit that. I know that you mentioned that you're dealing with lower involuntary attrition, and that was a part of it. So the question I have is just that versus seeking efficiency? What was the motivating factor to arrive at the level of workforce [indiscernible]?

Yes. So it's not one thing, it's multiple things, right? So we did -- and we talked about this already coming out of the third quarter, our attrition rates dropped dramatically. I don't think it was just us. I think it was more of an industry-wide event. And attrition for us has been our friend. We've been very, very successful in our advanced delivery initiatives to free up people and to reskill them and then put them back in into the business. But in order to put them back in, you need a seat. And when attrition is your friend and you -- and people are moving and turning over, you have plenty of seats from which to choose. When attrition dropped now there were no open seats, right? So look, it's part of it then is, well, we do have -- we do sort of have control of our own destiny. And we don't need this level of delivery resource, given how successful we've been at automation. So we'll just make an investment and we'll reduce our labor costs within the year, it's kind of neutral-ish at the bottom line, but next year it will deliver a big benefit to us. And then -- so that's part of it. Secondly, I'd say that we're always looking for ways to be more efficient. When we were born -- when we were spun out 4.5 years ago, it was a very heavy staff place. And we've been successful in reducing what it costs to run the place. Early on when we were on the IBM systems before we came out through the TSAs, we made the decision to go all cloud no mainframes. So we've taken big steps to improve how we operate, but there's always new ways to operate. There's always new ways to be more efficient. So part of this is also us operating more efficiently. So some of it is driven by attrition or lack of attrition and our ability to find new ways to deliver and very successfully implementing advanced delivery initiatives and some of it is just by our desire or need and our ability to cost less to run the place.

Staying with the workforce then Martin, I know it's -- I'm not looking for you to preannounce anything. And I know it's always a difficult subject to -- but just your thoughts on workforce longer term in general, and not even just AI, but everything we've talked about, there's so much change that's happening. And I know retraining and recertifying is a big part of the playbook. But as you're thinking around the workforce and where you want to be changed at all since last -- I guess, last year that you were here when I asked you the same question?

Look, it's -- we have a year now under our belts of understanding how, for instance, agentic can be used, right, in our own? And I mentioned we have 1,370 agents unique proprietary agents in the infrastructure. And as you know, they spin themselves up and they spend themselves down, so we could have 10,000 agents working. But those 10,000 agents are helping our teams collect the data, interpret the data, compare the data with all the data we have. So over time, everything that leads up to making a decision could change. But that's the most inefficient part of what we were doing anyway. The decision-making doesn't change. We still have a person over the top to make sure that as we adjust an infrastructure, as we learn about what's happening, that decision-making stays the same. It's still a person who has to make decisions. Now do we automate things? Yes, we automate things after. We're very comfortable that our engineering experts on a particular customer in a particular industry, any particular workload, are very comfortable letting the machine do some of this. But it's only the stuff that leads up to the decision that's changing, and it's making our teams far more efficient. I'll give you an example. One of our agents -- whatever our agents is -- and many of our agents are focused on root cause analysis, but one of our agent is very focused on when a problem occurs in the infrastructure, this agent, instead of having our engineers who used to have to do this, by the way, I should say, and this I'm sure it happens in JPMorgan the same way, if there's an incident in the infrastructure, everybody gets on the bridge and the half of the team that represents the infrastructure says the applications are the problem and they have that represent the applications, say, the infrastructure is the problem. That's every bridge call I've ever been on forever. That has gone away now because we actually have agents who are out now collecting all the logs from the relevant systems because they monitor everything. They're comparing those logs to what we've seen in other instances. So you get to root cause instead of in 30 minutes or 45 minutes or an hour, you get to root cause much, much quicker. But the engineers that sit on top of all this are still the one saying, okay, now that we have all this data, can we really focus on what the problem is, what's the course of action? So it is making us more efficient, but it's making us more efficient in the lead up to the decision the decision still needs the deep engineering expertise, the knowledge of the systems, the playbooks, the ways of working. All of that is still sitting within our teams. And I would say our teams feel like they're getting better supported now the AI because they actually have data faster and they can make progress faster. That, for us, allows our customers to think about can I move a little bit faster because I can get things -- I can fix problems faster, right? So it's sort of reinforcing to for our customers to know that our agents can help our experts get them going faster, get to the bottom of it. So they're willing to try a few more. They can try some new innovations. They can lean in a little bit on AI or some other new innovation.

So we have what 45 seconds left or so. Always trying to think a good closing question. I was debating a bunch of different ones. But we talked about a lot of different things. So the sector has been under a lot of pressure, a lot of questions around geopolitical, AI, everything you've talked about you're managing estates, it's much more stable. So what do you think -- what could change investor sentiment in your mind that you're excited about sort of giving us some data points to reveal and say, hey, things are actually moving in the direction that you planned?

Yes. The more time I spend with our customers, the clear it is to me that the role we play in getting them ready for the future and modernizing them is a very important role they can't do it without us, and we are in the trust business. So they do trust us to run their infrastructure for them, and they are increasingly trusting us to take them into the future and not just run their past. And again, I see that in the pipeline. I hear that from our customer base. We just signed just now a couple of weeks ago, I think we signed very big relationship with the Bank of Luxembourg. They put out a press release, we put out a press release. But I think it's emblematic of the way the important companies in industries like banking and insurance and all -- the companies that really make the world work, it's emblematic of how they think about what we can bring to the table and who should they partner with for all the complexity they're faced with. So I go through all of that because I think in the agentic world in the AI world, it's moving toward us. It's not moving away from us. in the AI world, in the agentic world, it's about your infrastructure, right, can the tracks support the speed of the train, it's about data, and that's what we do. That's what we've done for a long, long time. And again, we are the ones who help the -- who bring the stake to the sizzle, right? If today's sizzle is AI, we're the ones who are going to make that work.

Good. That's a good way to end the conversation. Martin, thank you for the time.

Thank you, Tien-Tsin.

Always fun.