Navigation

Investiment.io
Financial analysis platform

Microsoft Corporation

MSFT β€’ NasdaqGS

$517.99
β–Ό5.89 (1.12%)β€’ REGULAR

Microsoft Corporation

$517.99
β–Ό1.1%
πŸ“ŠSummaryπŸ“‹SEC FilingsπŸ‘₯Insider Trading
πŸ“„

MSFT Q1 2025 Earnings Call

October 30, 2024 at 12:00 AM

Company
MSFT
Quarter
Q1 2025
Date
October 30, 2024 at 12:00 AM
Speakers
2
Word Count
~8,647

Transcript Content

Operator
Our first question comes from Keith Weiss with Morgan Stanley. Please proceed.
Speaker
Congratulations on a really solid quarter. So, Satya, the expansion of capabilities, the speed of innovation, the magnitude of the opportunities ahead for generative AI makes this the most exciting period for software I've seen in my 25 years of covering this space. And based upon this call, it seems like you share that excitement. But in my investor conversations, that excitement also feeds two related questions, and they both have to do with constraints. The first is: what are the internal constraints or guardrails that Microsoft has when it comes to investing behind these innovations, particularly in relation to the funding of future generations of foundational models, where people are talking about price tags in the tens of billions or even $100 billion-plus? And then on the other side of the spectrum, what are the external constraints that Microsoft sees in building out this capacity to meet the demand and capture the opportunity, particularly constraints in your ability to power all these new data centers being built out and powered in an environmentally sustainable fashion? I'd love to get the Microsoft perspective on both those questions. Satya Nadella (CEO): Thank you, Keith, for those questions. I think on the first point, ultimately, when you think about, let's say, capital outlay for training because that's essentially what you're asking, it will be rate limited by your monetization of inference, given generation, right? So just like in the past, we would allocate capital to build out cloud based on the demand signal we were seeing, we would project the demand, and that's what we would build for. So, you can think of training essentially as that, right? which is you're building the next-generation model so that then you have a more capable model that drives more inference demand. Ultimately, even with all the scaling laws and what have you, I think you ultimately will normalize to having a pace. The best way to think about is given that Moore's Law effectively is working on the sort of silicon and system size. So it's just not compute, right? it’s efficiencies in compute, data as well as algorithms. You will want to sort of keep on that curve, which is you really want to refresh your fleet with Moore's Law every year and then effectively depreciate it over the period of its lifecycle. The inference demand will ultimately cover how much we invest in training because that's, at the end of the day, you're all subject to ultimately demand. So, the second piece of the external constraints, we have run into, obviously, lots of external constraints because this demand showed up pretty fast. If you think about even the most hit product of this generation, all are in our cloud, whether it's ChatGPT, Copilot, GitHub Copilot or even DAX Copilot; pick the top four or five products of this generation, they're all in and around our ecosystem. We ran into a set of constraints, which are everything because data centers don't get built overnight. There is power. So that's sort of been the short-term constraint. Even in Q2, for example, some of the demand issues we have or our ability to fulfill demand is because of, in fact, external third-party stuff that we leased moving up. That's the constraints we have. But in the long run, we do need effectively power and we need data centers. Some of these things are more long lead. I feel pretty good that going into the second half of even this fiscal year, that some of that supply/demand will match up.
Operator
The next question comes from the line of Brent Thill with Jefferies. Please proceed.
Speaker
Amy, good to hear the reacceleration in the back half for Azure. I guess many are asking, 34% growth in Q1 falling to low 30s. I know the comp is a couple of points harder, but is there anything else you're contemplating in that guide for Q2 to see that deceleration other than a tougher comp? Amy Hood (CFO): Thanks, Brent. Maybe this is a great question because I can sort of reiterate some of the points I made and tie them together a little bit. In Q1, the 34% in CC, as we talked about, that upside versus the 33% that we had guided to was primarily due to some revenue recognition benefits. I think about that on a sort of a pure consumption basis and AI as being 33%. You think about 1 point or 2 of deceleration that we've guided to, and the majority of that is due to, unfortunately, some supply pushouts that I mentioned and then Satya reiterated in terms of AI supply coming online that we counted on. The underlying consumption growth is stable Q1 to Q2. To your question on some ins and outs, it is certainly some ins and outs. I do, as you heard, have confidence as we get a good influx of supply across the second half of the year, particularly on the AI side, that we'll be better able to do some supply-demand matching and hence, while we're talking about acceleration in the back half. I'll also take the opportunity to say, when you see usage in AI workloads, we always intend to think about that as just a GPU exercise. The importance of having GPUs and CPUs be able to run these workloads is also important. So that's a piece of the acceleration in H2 as well.
Operator
The next question comes from the line of Mark Moerdler with Bernstein. Please proceed.
Speaker
Congratulations on the quarter. The question every investor obviously asks is a question on the CapEx growth and the CapEx spend. Obviously, half of that facility is an equivalent that have a longer life, but the other half is the rest of the components. Can you give any color on how you think of that growth? Does it return to the traditional approach, where basically CapEx is going to grow in line or slightly slower than cloud revenue? And if so, any sense of the timing? Do we have enough facilities online by sometime next year, et cetera? Any color would be appreciated. Amy Hood (CFO): Thanks, Mark. I think in some ways, it's helpful to go back to the cloud transition that we worked on over a decade ago, in the early stages. What you did see and you'll see us do in the same time is you have to build to meet demand. Unlike the cloud transition, we're doing it on a global basis in parallel as opposed to sequential, given the nature of the demand. As long as we continue to see that demand grow, you're right, the growth in CapEx will slow, and the revenue growth will increase. Those two things, to your point, get closer and closer together over time. The pace of that entirely depends on the pace of adoption. To Satya's point, some of that spend goes towards building the next training infrastructure so you won't see all of it in COGS. Some of it goes to OpEx when you're spending it on training. But in general, that's a healthy way to think about the balance as over time those do and should, like the last cycle, get closer together.
Operator
The next question comes from the line of Karl Keirstead with UBS. Please proceed.
Speaker
I'm actually not going to ask a question about the numbers, but Satya and Amy, I'd love to ask a question about OpenAI. Since the print three months ago, we investors have been hit with a torrent of media stories about OpenAI and Microsoft. I'd love to give Microsoft an opportunity to frame the relationship. It seems to me it's critically important. But we have been, I think, everyone on the line picking up signals that perhaps Microsoft wants to diversify somewhat at the model layer and offer customers choice. So, Satya, I'd love to get your framing of the relationship. And then in terms of the numbers, maybe this is a little bit more for you, Amy, but how does Microsoft manage the demands on CapEx from helping OpenAI with its scaling ambitions? How do you manage the impact on other income that you just gave us some color on? Satya Nadella (CEO): Sure. Thanks, Karl. I'd say first, the partnership for both sides, that's OpenAI and Microsoft, has been super beneficial. After all, we effectively sponsored one of the most highly valued private companies today when we invested in them and really took a bet on their innovation four, or five years ago. That has led to great success for Microsoft and for OpenAI. We continue to build on it, right? We serve them with world-class infrastructure on which they do their innovation in terms of models, on top of which we innovate on both the model layer with some of the post-training stuff we do as well as some of the small models we build. And then, of course, all of the product innovation. One of the things that my own sort of conviction of OpenAI and what they were doing came about when I started seeing products like GitHub Copilot as a product get built or DAX Copilot get built or M365 Copilot get built. We have a fantastic portfolio of innovation that we build on top of that. We feel very, very good about our investment stake in OpenAI. Our focus, and we're always in constant dialogue with them in a partnership like this where both sides have achieved mutual success at the pace we've achieved, means we need to push each other to do more to capture the moment, and that's what we plan to do. Amy Hood (CFO): And maybe to your other two questions, Karl, Listen, I'm thrilled with their success and need for supply from Azure and infrastructure and really what it has meant in terms of being able to also serve other customers for us. It's important that we continue to invest capital to meet not only their demand signal and needs for compute but also from our broader customers. That's partially why you've seen us committing the capital over the past few quarters, as it supports both grow together and for us to continue to grow the Azure platform for customers beyond them. I don't think of it as how do you balance it. We have customers who have needs and real use cases and delivering value today. If we can't meet that, we need to work to meet it, and that means working harder and faster to make sure we do that, which is what the team is committed to do. The second piece of your question on the impact to other income. Not to get too accounting-heavy on the earnings phone call, but I would say just a reminder, this is under the equity method, which means we just take our percentage of losses every quarter. Those losses, of course, are capped by the amount of investment we make, which we discussed in the Q this quarter as being $13 billion. Over time, that's just the constraint and a bit of a mechanical entry. I don't really think about managing that. That's the investment and acceleration that OpenAI is making in themselves and we take a percentage of that.
Operator
The next question comes from the line of Kash Rangan with Goldman Sachs. Please proceed.
Speaker
Satya, when you talked about the investment cycle, these models are getting larger and more expensive, but you also pointed out how the inference phase will likely get paid. How does that cycle look like an inference for Microsoft? Where are the products and the applications that will show up on the Microsoft P&L as a result of the inference rate of AI kicking in? Satya Nadella (CEO): Thanks, Kash. The good news for us is that we're not waiting for that inference to show up, right? If you think about the point, we even mentioned that this is going to be the fastest growth to $10 billion of any business in our history; it's all inference. One thing that may not be as evident is that we're not actually selling raw GPUs for other people to train. In fact, that's a business we turned away because we had so much demand on inference that we are not taking part in most of that. We are literally going to the real demand, which is in the enterprise space or our own products like GitHub Copilot or M365 Copilot. I feel the quality of our revenue is also pretty superior in that context. That gives us the conviction, even in Amy's previous answers, about our capital spend. If this was just about people training large models and that was all we got, we would ultimately still be waiting for someone to actually have demand. In our case, we have a diversified portfolio and are seeing real demand across all of that portfolio. Amy Hood (CFO): To add a little to what Satya is saying, what you're seeing is the $10 billion across inference and our apps is already what that momentum and that investment is doing and what that progress and revenue is what builds the next cycle of training. It's that circle as opposed to, oh, we're doing training and then inference. Much of the training investments that are fueling this revenue growth came before and we already funded that work. That's an important part.
Operator
The next question comes from the line of Mark Murphy with JPMorgan. Please proceed.
Speaker
I'm wondering if you can shed any more light on the nature of the supply limitations that you're mentioning that are impacting Azure in Q2, where that impact might be incrementally just a touch more than we expected? Is it more the GPU supply? Is there some element of power cooling or the ability to wire up the networks? Amy, should we infer that the supply is constraining Azure growth by roughly a couple of few points in Q2 or am I overestimating that? Amy Hood (CFO): To answer both those questions, Mark, I wouldn't think about it component logic in my Q2 answer. The supply pushout, as Satya said, was third parties that are delivering later than we had expected, that get pushed mainly into the second half of the year and, in general, Q3. So that's third parties where we have tended to buy supply inclusive of kits, so it's complete end-to-end third-party delivery. In terms of the impact, as I was saying, when you think about having flat consumption Q1 to Q2, there really are only two things that impact that difference and one was the help we got in Q1 from the revenue and accounting help, and then Q2 has been the supply pushout.
Operator
The next question comes from the line of Raimo Lenschow with Barclays. Please proceed.
Speaker
If we consider the current market, your early introduction of Copilot has positioned you well, and now we are discussing agents. Satya, how do you view this? It appears to me that we are evolving in our approach to productizing AI. How do you perceive the transition between Copilots, agents, and what may lie ahead? Satya Nadella (CEO): The system we built is Copilot, Copilot Studio, agents, and autonomous agents. You should think of that as the spectrum of things. Ultimately, the way we think about how this all comes together is you need humans to interface with AI. The UI layer for AI is Copilot. You can then use Copilot Studio to extend Copilot. For example, you want to connect it to your CRM system, to your office system, and to your HR system. You do that through Copilot Studio by building agents effectively. You also build autonomous agents. So, you can use even that as the announcement we made a couple of weeks ago is you can even use Copilot Studio to build autonomous agents. Now these autonomous agents are working independently, but from time to time, they need to raise an exception, right? Autonomous agents are not fully autonomous because at some point, they need to either notify someone or have someone input something. And when they need to do that, they need a UI layer and that's where again, it's Copilot. So, Copilot, Copilot agents built-in Copilot Studio, autonomous agents built-in Copilot Studio, that's the full system we think comes together, and we feel very, very good about the position. We are also taking the underlying system services across that stack that I just talked about, making it available in Azure. You have the raw infrastructure if you want it. You have the model layer independent of it. You have the AI app server in Azure AI. Everything is also a building block service in Azure for you to build. In fact, if you want to build everything that we have built in the Copilot stack, you can build it yourself using the AI platform. That's our strategy, and that's how it comes together. Brett Iversen (Head of Investor Relations): Operator, we have time for one last question.
Operator
And the last question will come from the line of Rishi Jaluria with RBC. Please proceed.
Speaker
I want to go and think a little bit about Copilot. How should we be thinking about the numbers here with the recategorization? It seems like that was maybe softer in the past than expected or maybe with the numbers this quarter starting to pick up. Can you walk us through what you're seeing on that? More importantly, how should we be thinking about your overall AI strategy on consumer versus enterprise, especially now with the stock on the fold? Satya Nadella (CEO): On the first part, we are very optimistic about the momentum we have with the commercial Copilot. As I mentioned earlier, this is the fastest growth we've seen for a new suite in M365. Compared to previous transitions like E3, E5, or the change from O to M, this growth is significantly quicker. The engagement from the Fortune 500 and their demand for more seats indicates strong momentum. Additionally, we want this to be a foundational change. It’s essential for users to implement security controls, deploy the system, and manage training and organizational changes. This isn't just a tool; it represents a new way of working, similar to the changes we experienced in the '90s with the rise of PCs. For instance, consider how business processes like forecasting evolved before and after the adoption of email and Excel; that’s the kind of transformation we expect with Copilot. Overall, we are pleased with our progress and market penetration. On the consumer side, we're excited to apply the same investments we are making in the commercial sector, where we see strong foundations. Notably, our revenue in search, news, and ads is growing faster than the market. Our consumer business, which generates over $10 billion, is experiencing significant growth that often goes unnoticed. We are confident in our plans for using AI in LinkedIn, which functions as a consumer business. Gaming is another area where these innovations will be implemented. Windows also plays a key role for us. We are particularly enthusiastic about integrating Copilot with PCs. It's crucial to move beyond a disconnected approach; we aim for hybrid AI that transforms the PC into the edge of AI, which is incredibly exciting for developers. We believe we are well positioned to advance with these investments. Amy Hood (CFO): Just to add one piece because I think Rishi, now that I'm listening and thinking through the question, it feels like you're wondering like, why am I not seeing Copilot if you've made all this progress and the results? The answer is, you already are. In that M365 commercial number, we've seen seat growth, but the majority of them are driven by frontline workers and small businesses. Those have a lower ARPU point and hence mask some of the ARPU we're already seeing not just from E5, but also this quarter from Copilot. As we go forward, that is where you'll see the impact will be in ARPU and M365 commercial; as Satya said, you'll see the impact of Copilot engagement across the same ex-TAC number. Brett Iversen (Head of Investor Relations): Thanks, Rishi. That wraps up the Q&A portion of today's earnings call. Thank you for joining us today, and we look forward to speaking with all of you again soon.
Operator
Thank you. This concludes today's conference. You may disconnect your lines at this time. Enjoy the rest of your day.