Microsoft Corporation (MSFT) Earnings Call Transcript & Summary

Okay. Good morning, everyone. I am Mark Murphy, Head of Software Research for JPMorgan. And it is a wonderful pleasure to be here this morning with Bill Duff, who is CVP and CFO for Microsoft Customer and Partner Solutions. Welcome to the conference. Thank you so much for making the long trip over here, Bill.

Thank you for having me, Mark.

Good to be here with you.

Yes.

Maybe you can begin with just a brief introduction of your background and your role at Microsoft for the benefit of the audience.

Yes. Thanks. So Microsoft Customer and Partner Solutions is basically our commercial sales organization. So I lead finance. I'm the CFO of that organization for about $200 billion of our revenue.

So when we run our CIO surveys, Microsoft has been in a very commanding #1 position. It's a large-scale CIO survey. The company has been cemented in what we call that CIO survey, Fab 5 list. We've had the company on our focus list. So we've been optimistic. When we ran our Microsoft partner survey for the March quarter, they landed 3% above plan. But I don't think we or anyone were really anticipating a squeaky clean quarter out of Microsoft. I don't think we were anticipating sitting here seeing 4 points of Azure acceleration because the environment just felt so uncertain. So very impressive performance out of Microsoft. Through your lens, Bill, what do you think surprised positively? And then especially within Azure, how did the stars end up aligning so well on that quarter?

Yes. Thanks, Mark. I got to say, internally, it never feels squeaky clean. There's a lot of work. We feel good about our overall execution on the sales side. And really, the demand that we saw in Q3 was strong across the board. We saw it in all geographies. We saw it in all segments from our largest customers down to our breadth customers or scale customers. I think what really surprised us, I guess, on the upside versus guidance was when we looked at our Azure business, there were a couple of things that broke our way and we're positive in the quarter. One is we saw our largest customers, we call them enterprise customers. We saw those customers accelerate. Particularly on the non-AI side, this is more of the core infrastructure side. We also saw our scale motions get better. We had called that out in Q2. We had made some adjustments in our execution in Q3, and we saw that improve a little bit. And then we were able to pull in some capacity on the AI side. And so all of those together really resulted, you combine that with the demand that we saw, and we were able to meet that demand and print a pretty good quarter.

So you -- it was part of what you mentioned there. Enterprise non-AI piece of Azure, we did walk away with that -- the impression that it was one of the parts of the business that performed materially above what -- even what you had expected going in, Bill. Is that the right way to think about one of the larger drivers there? And again, why do you think that -- why would that occur if business confidence seemed like it was slipping going into a trade war? Why do you think that, that piece would surprise positively?

It is the area. So we've been talking about our results as AI versus non-AI, but it's important to understand that those are very directly related. And so as customers invest in AI solutions, it pulls through infrastructure, whether it's on the compute side or on the data side. And so generally, we like to look at our overall Azure business and the growth in our Azure business is really a core signal of health in our enterprise customers. I think from a demand perspective, in speaking with customers, we're in areas that are focused on improving productivity and helping them with their operating leverage or managing costs. And so what we see a lot, particularly on the AI side, but also on the infrastructure side is just making sure that we have the right solutions to help them operate their business in a more efficient way. And certainly, we will be impacted by economic uncertainty and will be impacted by general economic conditions. But in the areas that we play and the sectors that we play in, we feel like we're pretty well positioned for investments in this type of environment where companies are really focused on operational efficiencies.

Okay. So AI pull-through and other factors you mentioned. The commercial bookings, Bill, was also a great performance. And I think you're racking up a bunch of quarters in a row where we see this double-digit cadence on the commercial bookings growth. From your vantage point, what do you think it is? Anything in particular you'd call out that is driving that bookings performance?

Yes. Bookings is something we look at closely, and it's a broad indication of companies' confidence in Microsoft for the future. And so that's really how we see it. We are invested in some of the most strategic areas as these companies try to transform their business, whether it's on the AI infrastructure side or even just our general Azure business. We feel like that this is a signal of the confidence in Microsoft and our long-term road map and the products and services that we deliver. And we look at our -- as this as the way that our customers really come along with us, and we make -- we can make sure that we support them in their aspirations as they transform their own businesses. So we look at this as a real positive.

So you have this incredible vantage point where you're overseeing the global commercial business. You're overseeing that from SMB all the way up to enterprise, and it includes the public sector as well. I think, Bill, it puts you in a very advantageous position to comment on the tone and tenor of customer behavior amidst all the headlines that we're seeing, which are changing on a daily basis, that changed quite a bit yesterday. So we have economic uncertainty. We have quite a bit of excitement about AI. What are the signals that you are seeing about this kind of business willingness to invest at the moment?

Yes. The signals that we're seeing, particularly on the AI side, are pretty positive. It seems like every organization out there and certainly every large organization is investing and experimenting with different solutions to make their company more efficient or to drive top line revenue growth. And there's a ton of examples that we have. We talked a little bit about these in our latest earnings release. But for example, Dow is using AI to help with their overall supply chain. And so they get 100,000 Acrobat files for invoices, and they make sense of this and they really try to make their supply chain more efficient. They're seeing savings of millions of dollars. Fujitsu is another example. This is more on the revenue side. They're deploying an agent that's a sales agent. And so what the sales agent does is it provides their sales folks with detailed information about different customer prospects that pulls together information and really helps them close deals. They're seeing increased Salesforce productivity for that. And it's also helping them to onboard new sales leaders into the organization. And so you don't necessarily have to have the years and years of experience with these various customers, you can come up to speed and be more productive much more quickly. And those are the kinds of examples that we're seeing on the AI side, in particular. I think people are really looking at very pointed solutions. They see business problems. They see strong ROI around a lot of these types of projects. And so they're investing in them because they see a pretty quick return.

Okay. So those are fascinating examples because it's real-life deployment of Agentic AI technologies happening out there in the wild with some pretty mainstay organizations. So I think it's great to hear that. In the background, though, we've had this optimization intensity bill. That was something that started to settle into the environment in the second half of 2022. Do you think that this optimization intensity is -- do you think it's fading off? Do you think that it is kind of post the peak of those headwinds where companies are trying -- they're trying to purge, they're trying to rationalize the licenses. Where do you think we are in the cycle?

Optimization is just part of the life cycle. What we see with customers is they'll start a new project, they'll invest in that project, they'll get it up and running, then they'll start optimizing it, making sure it's as efficient as possible and then they'll reinvest those savings in the next project. And so we don't necessarily see it as these big cycles. We see it more on a project-by-project basis. And so we're not seeing any signals that things are out of the ordinary at all on that front. We just think it's part of the normal process and just the maturation of how companies are looking at these investments and want to make sure that they are running them as efficiently as possible and then making sure that they have the opportunity to invest in the next generation of products and solutions.

So one other question, just thinking back to the earnings reports, Bill, because we -- this is a question we were getting from investors. Philosophically, why not incorporate some cushion into the guidance? And this is a question we can apply across a lot of the software industry right now. But just to account for the possibility that launching the trade war and some of the [indiscernible]efforts could have a lagging effect and cause a slowdown. And again, like I said, the headlines are changing on this every day. But we've been through a month where air freight is down, shipping freight is down. We've had -- airlines have been missing forecast, retailers have been missing forecast. How do you think that through?

From a guidance perspective, we've taken a pretty consistent approach. We look at -- as you mentioned, we have broad exposure across many different geos, across many different customer segments, across many different industries. And we just look at the signals that we are seeing. And that's generally how we set forecast. Our goal is to be closest to the pin. And from the signals that we're seeing, that's really how we established our forecast or our guidance that we talked about in the earnings call for our Q4. And as I think we mentioned on the earnings call, the signals -- we saw the signals being consistent through April. And so our goal is to get closest to pin Of course, if things change and if those signals change and if things -- if the economy moves in a different direction, those signals might change, and we would have to update. But for right now, we feel pretty good about the signals that we're seeing and the demand that we're seeing out there for our products and services.

Yes. The April comment was a big one because I think it's a little uncharacteristic that Microsoft would comment on 1/3 of a quarter like that. But I want to go back to -- because you gave us some great examples a moment ago of what some massive companies are doing in the realm of AI and Agentic. When we run, again, coming back to our large-scale CIO survey, there were 4 products that are -- they're kind of way up in the upper right for CIOs are aware of it and they're impressed. They all relate to Microsoft and OpenAI. And so those products, of course, it's ChatGPT itself. You have Azure OpenAI service. It was M365 Copilot and then it was also GitHub Copilot in those top 4 positions. But then we step back and we think about it, the tapestry that Microsoft has is expanding so rapidly. There's so much happening. And you actually mentioned -- I don't know if you called it sales agent or Sales Copilot.

Sales agent.

Sales agent. There's a Security Copilot and Security Agent, you have Copilot Studio. There's this whole broadening lineup of agents that you have. Can you give us a little bit of a window into what is resonating there within the Microsoft AI offerings? Yes. In terms of -- and I mean it in terms of bookings momentum.

Yes. And we're seeing pretty strong momentum across the portfolio. M365 is a big bet. M365 Copilot is a big bet for us. And what's really interesting is sort of what we see on the life cycle of M365 Copilot. And so what we see is a company will generally purchase seats, kind of understand how it works and then come back for more once they get comfortable with it, once they see the productivity enhancements for their employees and once they see the ROI of M365 Copilot in particular, and then they come back and they buy more. And so in our latest earnings call, we talked about 3x year-over-year usage growth. And so we're seeing very, very strong usage growth. We're also seeing more and more customers. I think we had a record number of customers in Q3 come back to us and expand their offerings. And so most people or most big companies are experimenting. And then the question is, is once they get to a certain comfort level, certain level of usage, that's where they expand. And what's kind of interesting is as a user of M365 Copilot, that's been my experience as well. And we were talking a little bit prior to getting on stage and I've been at Microsoft a long time, so I've been at Microsoft over 20 years. But this role for me is a relatively new role. I got this role within the last 6 -- a little bit over 6 months ago on the commercial sales side. And I use Copilot, M365 Copilot on a daily basis. It's kind of amazing when I look back and when I first adopted M365 Copilot, I was using it as kind of a search engine. It was kind of a very, very powerful search engine for me to find data on different information or decks or things like that. And now I'm using it and incorporating it on a day-to-day basis. It's made me more productive. I have M365 Copilot up on the screen at all times. I'm constantly interacting with it. And so I think this pattern isn't unusual. And so whereas before, I used it for search or something like that or maybe to look up at acronym at Microsoft, but now I use it to understand and analysis or even some of our more experimental stuff. We have researcher in preview. We have analysts in preview. These are really powerful tools that are coming on board. So M365 Copilot is getting more and more powerful, and I'm using it more and more in my daily interaction. And just from that personal experience, I think getting back to customers, I think that's how they see it as well. Like there's a lot of excitement. And when you use it on a daily basis, you can really see the power and you really understand how much you rely on it, where in the past, you would have to make a phone call or reach out to somebody else to get a piece of information or a deck or synthesize something. I'm doing that a lot myself on a day-to-day basis with M365 Copilot.

And I want to come back to that in just a moment because we do hear and we do see a range of opinions and assessments on that specific product. I want to come back to it. But just before we do, I want to ask you AI adoption at the infrastructure layer compared to at the apps layer. You had -- Microsoft had quickly gotten to over $13 billion in AI services ARR. That was as of December. We think you probably added another $2 billion or $3 billion in March. So there's no company close to this scale that we're aware of. It's growing well over 100%. And then, Bill, we compare that to the landscape of SaaS software providers, they're struggling to get their AI products to get -- to reach 1 percentage of revenue or bookings. So can you compare what's happening there, AI in the apps layer, at the infrastructure layer? And why do they seem to be moving at different speeds?

Well, I don't think they're moving at different speeds. Obviously, the AI infrastructure demand and growth is coming from applications, and it's coming from either internal applications that companies are building or end-user applications that are using This technology. And so I think you're seeing a tremendous amount of pace and accelerating functionality across both the infrastructure layer and everything we're doing to build out this infrastructure to make it easy for customers to use this infrastructure. And so making sure that each layer is the most [indiscernible]as possible. We're investing a lot in Fabric, which is the data analytics layer to make sure that the data is able to be used in a way by all of these agents and all of these new applications. And then at the foundry layer, we're really making it very, very easy for developers to build these applications overall to help their businesses perform better. And so I think you're seeing this really symbiotic relationship between applications and infrastructure. On the app side, I think there's so much innovation that's happening. And it's probably just a little bit more difficult to pinpoint exactly what is AI versus non-AI. I mean AI is just an underlying infrastructure. You can think of -- I've heard it described as something similar to electricity or other types of foundational technologies. And so one of the ways to think about that is you are starting to see it show up. There's been a lot of companies that have talked about how AI is improving their ad models and their ad revenue. And so that's clearly a part of the end user application and the innovation that's happening there. And I think it's just being embedded across all sorts of productivity and SaaS applications. So the concept of a productivity or a SaaS application without AI is quickly going to go away, and it will be so just fundamentally embedded in all of these applications and services. And I think you're seeing that.

So it's happening everywhere and it's symbiotic. Let's come back to -- I want to come back to M365 Copilot. Like I said, there's a range of opinions. Frankly, I think institutional investors are a little lukewarm on M365 Copilot -- now when we survey CIOs and when we survey the Microsoft partner ecosystem, they're not lukewarm on it. They're leaning in, they're deploying it. They're seeing success. They're seeing hours saved, et cetera. And they do see these repeat purchases. They're pretty confident in it. What are the learnings here from M365 Copilot? And why do you think we have that range of opinions?

I don't know why we have the range of opinions. I do think that customers are clearly the ground truth on this. And so when we talk to our customers, as I mentioned a little bit earlier, they're excited about the opportunity. They want to make sure that they're comfortable with how they're going to use it. And so what they're doing is they're deploying it to a select generally a select group of individuals. Generally, they'll look at a department or something like that, they'll deploy it. They'll track usage. They'll make sure that they have the proper governance in place around data. These are very, very powerful tools. And so you want to make sure that your infrastructure is set up properly so that you can reason against the data in the right way and expose the data in the right way to certain users. And so I think companies are both getting more comfortable on the infrastructure side, making sure that their state is ready for this technology and then they're watching usage. And I think the ROI, what they're telling us and what they're seeing is the ROI is there. And so they're looking to expand. And so I think that's the right way to think about it is broad adoption across many, many different companies, but at a relatively small number of seats and then grow from there.

Okay. Okay. So then it's early and there's a runway. Let's talk about -- I want to ask you for a moment about potential for AI model convergence. Again, this large-scale CIO survey that we run, it looked to us, every time we run it, there's a disproportionate leadership position for the combination of Microsoft and OpenAI. And so we saw CIOs are planning to do 68% of their work on Microsoft and OpenAI and 17% that we were going to do on Amazon, 12% of that work that we're going to do on Google. So we look at it and we think, well, they see differentiation in the products. But then we've heard Satya Nadella say multiple times that the AI models will increasingly converge, and you've had -- there's been a competitive response you've got Anthropic, you've got Google, you've got [indiscernible] you've got DeepSeek, you've got others. How do you plan to maintain a differentiation in the AI market if those LLMs themselves are going to head towards equivalency or convergence?

There's been a tremendous amount of innovation on the model side. And certainly, we expect that to continue. And it's great to have a vibrant competitive environment. But we think about the competitive landscape on the -- and the core differentiation broader than just models and on the infrastructure side. And so this is something that Microsoft has been doing for a couple of years, really thinking about Azure and thinking about Azure as the infrastructure for AI in the future and what do you need to do there? And so we've been innovating across every layer, making sure that our foundational technology is as efficient as possible so that we can support all these applications. I talked about fabric in the data and AI layer and making sure that we can operate in very complex environments where people have many different technologies. It's not like everybody is sort of creating a new environment that's a pristine environment for AI. Every company is coming from their own legacy infrastructure, whether it be still something on-prem, but also something in the cloud. And so when we architect Azure, it is to make sure we can operate in that complex environment, and we're being as efficient as possible along every layer. Same thing on the foundry side, really making sure that we have an opportunity to bring this technology to developers so that AI factory, AI app factory is what we call it externally. And so when you think about making applications in an AI environment, how do you make it as seamless, as easy as possible so that these companies can get up and running and developers can get up and running in this area. And so we think there are big opportunities for differentiation on those different layers of our overall infrastructure stack.

Okay. So it's a much broader build than just the model, and it's all up and down all the layers of that infrastructure.

Even to the data and compliance side and security side and everything else, it really is a broad opportunity for differentiation for us.

So let me ask you about on that exact topic. I think investors have noticed there's been a deeper level of initiatives recently that I would call relating to cross-cloud interoperability. One of these has been -- that's been noteworthy has been Oracle Database at Azure. But there are also a number of -- part of what you -- I think you just alluded to, you have Azure Arc itself, right, which can make other clouds look like they're running natively in Azure. You have things like Azure Sentinel is running cross-cloud these days. Can you help us understand what is happening with all of these initiatives?

Yes. I think the way to reason against it or the way to think about it from a Microsoft perspective is we have a lot of customers that have a lot of different environments. It's a complex space. And so we try to meet customers where they are. We are -- we've been a partner-focused and a platform-focused company Since our inception. And so we're very comfortable in that type of environment, working with partners across the spectrum. And you can see this in our business overall, whether it's our partnership and the way we work with Oracle environments or other companies as well. We just try to meet customers where they are. We understand not everybody has the same starting place. We understand that there's a lot of different reasons why different companies have chosen different infrastructure providers for certain elements of their infrastructure. And our goal is to make our customers' entire infrastructure work better, and we're going to do everything we can to make sure we interoperate well, but also provide a path to the future where we think that the technologies that we bring are going to be really important and valuable as these companies go through these transformations.

Okay. Let's touch on security because just a week or 2 back, we had met with the heads of your security business [indiscernible ] Security Conference. And we were thinking back on this. I think the last update that we could recall of the security business was a little over 2 years ago, it was January of '23. That business had reached $20 billion plus, and it was growing more than 30%. So there's a lot of scale. There's a lot of trajectory behind that business. And it's interesting to think about where it could be a couple of years later. Can you help share what the security conversation looks like with Microsoft's customers today?

Yes. Security is obviously paramount to our customers, and it's very important to us. It's complex and very, very dynamic environment. There's a lot of players. A lot of our customers have multiple security vendors that they work with to ensure that their infrastructure, their app layer and everything operates in a secure and compliant manner. And so it's an important market for us to be in. Our strategy and our North Star is really around consolidation, making sure that we have the products and services in place that can look across our customers' estate on the infrastructure side, on the app side. We think we're extremely well positioned to have a differentiated solution because of that simplicity of bringing everything together and because of the signals that we see because we're a large player on both the productivity and application space as well as the overall infrastructure space. The signals that we get, that we incorporate in our solutions are truly differentiated. But our strategy is consolidation. Our strategy is simplification and providing a holistic solution to our customers. And we feel like we've done pretty well there. Of course, there's always more to do, but we're continuing to make progress. We're leaders in over 20 different workloads on the security side. And so we have very, very broad adoption of our security solutions. I think there's a really interesting opportunity in security, especially as you go into the Agentic world. And I touched on this a little bit before with M365. But as you go into an Agentic world, think about whether it's access, identity and access to information and you have an agent working on working on behalf or working actually autonomously. And so what access are you going to give that agent? How are you going to manage that agent? You can think about it in terms of data governance and making sure that your data governance solutions are there. So when we look at the opportunity and especially with this transformation with AI, we see a big opportunity for us in the security space. So it's a critical market for us. We think it's going to be an interesting market. We think we have a differentiated solution. So we're pretty excited.

It's like onboarding an employee literally when you bring those kinds of agents in. Okay. So in the last couple of minutes here, I want to try to do 2 quick ones. And I want to try to connect 2 observations that we have had. So one is on the models themselves, the AI models. They're reaching a PhD level of understanding. The code generators are resonating with developers. The way that they're -- the adoption in contact centers, they've got these very high solve rates. You see all this happening in the AI realm. And then I look at Microsoft's headcount growth, and it was flat in fiscal '23. It only grew 3% in fiscal '24. So repeatedly, I think we see this, Bill, there's minimal headcount growth and then there's healthy double-digit capEx growth. And it's a little different structure. Is your own use of AI products kind of contributing to the lower headcount growth? And if so, can you speak to the stronger use cases you see internally?

Yes. I think operating leverage is something we focus on a lot at Microsoft. So it's broader than AI for sure. And so we spend a lot of attention figuring out what are the priority spaces for us to invest in, and we're constantly readjusting our portfolio to make sure that we just don't continue to invest in legacy products and services. We're focused on the future, and we make sure that our resources are focused on the future. So a lot of this is around how do you balance your workforce and make sure we're working on the things that we think are the biggest TAM opportunities, places where we're differentiated and places where we have opportunities in the future. We also think about org health as well. And so we think about layers of our organization. And so what I mean to say by that is we have the sort of traditional ways of just thinking about how to make sure that we're focused on the right opportunities and we're investing in the right places across the company. That said, we do see big for AI leverage in our company. I mean some of the examples, we're using it in our support organization. We're saving hundreds of millions of dollars a year just in terms of enabling customers to support the products by themselves. You don't need human interaction, also to make our support reps more productive. We use sales agent internally to help our sales representatives get more information and be more productive when they're talking to customers, less time on back office minutia, more time with customers out there. On the finance side, we use it deeply in compliance. We have AI looking at all of our complex big deals and making sure that we're compliant and operating in a compliant way. We're using it in marketing company. So -- and this is also on the developer side, which you heard Satya talk about recently. And so we're using it deeply across everything. And so but it's more about the productivity angle, and it's more about making sure that we have a good employee experience and that everybody is able to support what they need to support to grow our business.

You've been way ahead of the curve with it. So Bill, in the final 30 seconds, I want to try to ask you before we let you off the stage here, what do you think is one thing everyone in this room, which is a pretty large room, will be talking about a year from now.

And the pace of innovation is incredible. We're seeing it at Microsoft. We're seeing it with our customers. This stuff is happening real time. And I think this trend around agents and Agentic action and even autonomous Agentic action is going to be a big trend over the next year. And I think a year from now, we're going to look back and say we've made more than a year of progress. It's kind of incredible what we're seeing in terms of pace, and I think that's going to continue and even accelerate.

It's moving at light speed. So we can't wait to see, Bill, that level of innovation that's going to be coming forward for the next year. Thank you so much for taking the time to fly across the country and be with us here today.

Thanks, Mark.