Microsoft Corp ($MSFT) Q4 2025 Earnings Call

Thank you, Jonathan. It was a very strong close to what was a record fiscal year for us. All up, Microsoft Cloud surpassed $168 billion in annual revenue, up 23%. The rate of innovation and the speed of diffusion is unlike anything we have seen. To that end, we are building the most comprehensive suite of AI products and tech stack at massive scale. And to provide more context, I want to walk up the stack starting with Azure. Azure surpassed $75 billion in annual revenue, up 34% driven by growth across all workloads. We continue to lead the AI infrastructure wave and took share every quarter this year. We opened new DCs across six continents and now have over 400 data centers across 70 regions more than any other cloud provider. There is a lot of talk in the industry about building the first gigawatt and multi-gigawatt data centers. We stood up more than two gigawatts of new capacity over the past 12 months alone, and we continue to scale our own data center capacity faster than any other competitor. Every Azure region is now AI first. All of our regions can now support liquid cooling increasing the fungibility and the flexibility of our fleet. And we are driving and riding a set of compounding S curves across silicon, systems, and models to continuously improve efficiency and performance for our customers. Take for example GPT-4o family of models, which have the highest volume of inference tokens. Through software optimizations alone, we are delivering 90% more tokens for the same GPU compared to a year ago. Beyond the AI fleet, we continue to build our commercial cloud to address customers' unique data residency and sovereignty requirements. This quarter, we introduced the Microsoft Sovereign Cloud, the industry's most comprehensive solution spanning both public and private cloud deployments. All of this innovation is driving our strong results. We saw accelerating growth from migrations again this quarter. Nestle, for example, migrated more than 200 SAP instances, 10,000-plus servers, 1.2 petabytes of data to Azure with near-zero business disruption. That makes it one of the largest and most successful migrations in business history. The next big accelerator in the cloud will be quantum, and I'm excited about our progress. In fact, earlier this month, we announced the world's first operational deployment of a level two quantum computer in partnership with Atom Computing. This is how we will continue to think and make investments with decade-long arcs while making progress every quarter. The next layer is data, which is foundational to every AI application. Microsoft Fabric is becoming the complete data and analytics platform for the AI era spanning everything from SQL to no-SQL to analytics workloads. It continues to gain momentum with revenue up 55% year-over-year and over 25,000 customers. It's the fastest-growing database product in our history. Fabric OneLake spans all databases and clouds, including semantic models from Power BI, and therefore it is the best source of knowledge and grounding for AI applications and context engineering. Azure Databricks and Snowflake on Azure both accelerated as well. Cosmos DB and Azure PostgreSQL are both powering mission-critical workloads at scale. OpenAI, for example, uses Cosmos DB in the hot path of every ChatGPT interaction storing chat history, user profiles, and conversational state. And Azure PostgreSQL stores metadata critical to the operation of ChatGPT as well as OpenAI's developer APIs. This year, we launched Azure AI Foundry to help customers design, customize, and manage AI applications and agents at scale. Foundry features best-in-class tooling, management, observability, and built-in controls for trustworthy AI. Customers increasingly want to use multiple AI models to meet their specific performance, cost, and use case requirements, and with Foundry, they can provision inferencing throughput once and apply it across more models than any other hyperscaler, including models from OpenAI, DeepMind, Meta, XAI's Grok, and very soon BlackForestLabs and Mistral AI. We sim-shipped 15 models from OpenAI alone on Foundry this year, providing same-day access to state-of-the-art models deeply integrated with our infrastructure and tools. And we are seeing accelerated adoption of our new Foundry agent service, which is now being used by 14,000 customers to build agents that automate complex tasks. For example, Nasdaq is using Foundry to build agents that help customers prepare for board meetings, cutting prep time by up to 25%. All up, 80% of Fortune 500 already use Foundry, and when we look narrowly at just the number of tokens served by Foundry APIs, we processed over 500 trillion this year, up over 7x. This is a good indicator of true platform diffusion beyond a few head apps and services. Talking about the app layer, these applications are becoming embedded in our daily work and life. Our family of Copilot apps has surpassed 100 million monthly active users across commercial and consumer. And when you take a broader look at the engagement of AI features across our products, we have over 800 million monthly active users. Microsoft 365 Copilot is becoming the new way to organize work and workflow and work artifacts. We rolled out our biggest update to Microsoft 365 Copilot to date this quarter, bringing together chat, search, create notebooks, as well as agents into one intuitive scaffolding. With this innovation and continued product improvements, we are seeing real momentum. Customers continue to adopt Copilot at a faster rate than any other new Microsoft 365 suite with strong usage intensity as shown by our week-over-week retention. And we saw the largest quarter of seat adds since launch with a record number of customers returning to buy more seats. Barclays, for example, will roll out Microsoft 365 Copilot to 100,000 employees globally following a successful initial deployment of 15,000. UBS is expanding its deployment to all of its employees after initially rolling it out to 55,000 of them. And Adobe, KPMG, Pfizer, Wells Fargo all purchased over 25,000 seats this quarter. Tens of thousands of organizations have already used our researcher and analyst deep reasoning agents in the first weeks of availability. And we have introduced group level agents in teams like facilitator and interpreter, which generate real-time translation and notes in meetings. Hundreds of partners like Adobe, SAP, ServiceNow, and Workday have built their own third-party agents that integrate with Copilot in Teams. We're also seeing more customers use Copilot Studio to extend Microsoft 365 Copilot and build their own agents. This year, customers created 3 million agents using SharePoint and Copilot Studio. And with Copilot tuning, they can easily create agents fine-tuned on their company's data, workflow, and style that reflect their unique tone, language, and expertise. We're also seeing great traction among specific roles and functions starting with developers. GitHub Copilot continues to have great momentum in IDE with agent mode and new form factors like coding agent, which is capable of asynchronously executing developer tasks. We have 20 million GitHub Copilot users. GitHub Copilot Enterprise customers increased 75% quarter over quarter as companies tailor Copilot to their own code bases. And 90% of the Fortune 100 now use GitHub Copilot. More broadly, GitHub usage and repos are seeing explosive growth because of AI. AI projects on GitHub more than doubled over the last year. The surge in vibe coding projects and AI coding agents, whether it is Claude Code, Codex, Cursor, or GitHub Copilot, are generating more pull requests and more repos on GitHub. And our code review agent is being used heavily across the platform, performing millions of code reviews each month. In healthcare, we had a breakout year for Dragon Copilot. Customers used our ambient AI solutions to document over 13 million physician-patient encounters this quarter, up nearly 7x year over year. For example, at Mercy Health System, more than 1,000 physicians are already using Copilot to reduce administrative burden so that they can focus on providing better care. They have saved more than 100,000 hours to date and plan to expand to all 5,000 providers. As one physician put it, best thing to happen to my practice in 10 years. And in security, we were the first in the industry to introduce agents to help defenders autonomously handle high-volume security and IT tasks. More broadly, AI is driving a fundamental change in the BizApps market as customers shift from legacy systems to agentic business applications. Dynamics 365 took share this year, and we're winning customers in every industry like Verizon with sales, Domino's Pizza the group with ERP, 1-800-Flowers with contact center. When it comes to consumer apps, we're innovating across all surfaces. In fact, on Monday, we introduced Copilot mode in Edge. It's especially exciting to see the innovation coming back to browsers. Copilot mode brings together Copilot composer, chat, discover, search, and actions to build the next generation of browser for the AI age. Our Copilot consumer app also continue to see strong growth and engagement and successful sessions. And we're bringing Copilot to every Windows 11 PC. With Copilot vision, you can share your screen with Copilot and get real-time insights and assistance on anything. And we're well-positioned as we approach Windows 10 end of support in October thanks to Windows 11 and Copilot + PCs, which offer customers compelling security as well as AI value. Talking about security, it underlies our cloud and AI infrastructure as well as our Copilots and agents. We have launched over 100 new capabilities over the past year. Just last week, we added a modern data lake to our SIM, Microsoft Sentinel, bringing together customer data across our first-party tools as well as third-party ecosystem over 350 connectors. We're also extending the systems customers already use for governance, identity, security, and management to protect every AI agent. Entra now extends identity permissions, policies, and access controls to agents. Defender secures nearly 2 million generative AI apps. Purview is used by three-quarters of the Microsoft 365 Copilot customers to protect their data. And all up, we now have nearly 1.5 million security customers and continue to take share across all major categories we serve. Before I wrap, I want to talk about two consumer businesses of ours with massive end-user reach, LinkedIn and Xbox. LinkedIn is home to 1.2 billion members with four consecutive years of double-digit member growth. All up, comments on LinkedIn rose over 30% and video uploads increased over 20% this year. We continue to bring AI to every part of LinkedIn experience, introducing agents across hiring as well as sales. When it comes to gaming, we have 500 million monthly active users across platforms and devices. We are now the top publisher on both Xbox and PlayStation this quarter with successful launches of Forza Horizon 5 and Oblivion Remastered. The Call of Duty franchise has never been stronger. 50 million people have played Black Ops 6. Total hours surpassed 2 billion. Minecraft saw record monthly active usage and revenue this quarter thanks in large part to the success of the Minecraft movie. And we have nearly 40 games in development, so much, much more to come. We surpassed over 500 million hours of gameplay streamed via the cloud this year, and Game Pass annual revenue was nearly $5 billion for the first time. In closing, we're going through a generational tech shift with AI, and I have never been more confident in Microsoft's opportunity to drive long-term growth and define what the future looks like. With that, let me turn it over to Amy to walk through our financial results as well as the outlook.

So, these are workload results that are invaluable for us to learn to build both the products as well as the platform. And then broadly, you know, they or rather over time they will be broad diffusion. In fact, one of the things that Amy and I track is not just the head app usage, but also what's the sort of all the tier two applications that are being built.

So, that's sort of that speaks a little bit Keith to I think your question is as long as we have head apps shaping the platform and then after that we have the broad diffusion happen, which in some sense we're both of those is what we're seeing. So, I feel very good about being in decent standing going forward. Thanks, Keith.

Yeah, I'll start and Amy should feel free to add. I mean, if I just broaden out beyond just SaaS as a category, I think just like, you know, the server to cloud transition was an expansion of essentially usage of servers. That is essentially what happened with the cloud, right? Which is we did a bunch of servers except that the expertise required, the capital required, the time required to bring up servers, build it out, scale it was just all hard and so therefore the market was a certain size, whereas with the cloud you could sort of buy it with flexibility. You could burst and you could spin up and spin down. The expertise required came down, so it was just, you know, orders of magnitude. That's what's happening, you know? So, if you sort of even subscribe to this point of view that intelligence is basically log of compute, that means compute's going to grow and you've got to use it as efficiently as possible to just keep creating intelligence.

Now, how does it manifest beyond just the infrastructure? I kind of, you know, to Keith's earlier question, I talked a little bit about how infrastructure is getting shaped, data layer is getting shaped, the app server is getting built. These are all classic categories of infrastructure that will continue, but they will be an order or two of magnitude more. So, literally like in fact, one of the other things we track is every GPU requires storage and compute. That ratio is another thing that is really exponential for infrastructure growth. So, when you go to the app layer, the SaaS apps themselves are now building in effectively agentic and chat interfaces with intelligence and they're also building autonomous agents. Agents are kind of like applications like a database application, perhaps, but they are being used increasingly in applica- you know, inside of a user interaction. I think a good example is GitHub Copilot. It got started as an you know, code completions on an IDE. Then we added the chat interface to it. Then we added the agent mode to it. And now we have an autonomous agent, which in fact works completely asynchronously, right? So, all those four things are now part of essentially GitHub. And by the way, it also turns out that every other tool that is also doing any form of coding is adding more and more GitHub repos. So, if I have to think about GitHub monetization, we have an opportunity around just monetizing GitHub Enterprise. And then we have the ability to think about GitHub Copilot. And GitHub Copilot as with all these form factors. And so, that's exactly the same thing that's happening with Microsoft 365. That's the same thing that's happening with Dynamics 365. So, you have to be very open to taking your data tier, your business logic tier, and your UI tier, and sort of being more expansive in it. As long as you do that, it's just that usage goes up. And that's what I think shows up in the results.

Yeah, I mean, just there are three things that are really happening. One is the migrations. A good example would be what I referenced in my remarks with Nestle with the SAP instances they moved along with a lot of the data that's associated with it and a bunch of servers. So, that's kind of a classic example, I think, whether it's VMware migrations or migrations of SAP or even just our own server migrations, they're pretty healthy. And as I say, you know, it turns out that we're still not anywhere close to the finish line. If at best, maybe in the middle innings of that. The second thing that's also happening is cloud-native applications that are scaling. This is even excluding all of the AI stuff. Just the classic cloud-native e-commerce company, let's say. These are scaling in a big way. And some of those customers were not on Azure previously, but now they're increasingly there because they've come for AI, perhaps, but they now stay for more than AI. And so, to me, that's another thing you see in overall what's happening across the Azure number. And then, of course, there are the new AI workloads. So, those are three things that are all in some sense building on each other, but that's kind of what's driving our growth.

Yeah, I don't know if anything really surprised us, but I think what we're noticing in our own build-out of these AI applications and in general is the platform's becoming more than here's a model and here's an API, you know, make some calls, right? I mean, that's in some sense was a bit of the state of the art, maybe even a year ago. Whereas now, you have essentially these very stateful app patterns that are emerging that require quite a bit of rethinking of even the app stack. I mean, take even the storage tier stuff, right? The degree of sophistication you have in hey, how much of an index do you really want to build by preprocessing so that your prompt engineering or context engineering, as I call it, can be better and higher quality. So, I think all of that is emerging. So, when I look at a product like Azure search, fabric, and Cosmos DB, all of the things, the frameworks around it are just becoming robust to build serious applications.

And so, that's what I feel great about is the learning curve inside the company, outside the company, the diffusion of the stack, the speed with which that's emerging, that you can build applications is much faster, right? You know, I always go back and say, hey, when a relational database came out, they took a while for people to build an ERP system, let's say. And this thing, we're kind of building pretty sophisticated applications at a very, very fast clip based on the degree of maturity that's emerging.

Yeah, I think that's right. Even inside of Copilot, I'm sure you're seeing it, right? You now have analysts and researchers, to just talk about two examples. And of course, people all the third-party agents. So, yeah, there is a lot more of this just not request-response. It's about spawning essentially applications that then go do work and come back. But the UI still remains very important, right? Even for asynchronous work, to instruct the asynchronous work, you need UI. And to monitor asynchronous work, you need UI. It may be different. It may not be a chat interface. And of course, you need a way to then inspect what the asynchronous work is, right? So, even take the example I was giving on GitHub. Even if you're not using GitHub Copilot to create the code check-in or the pull request, interestingly enough, we're seeing massive increase to GitHub Copilot code review agent, even if you used maybe Claude code or whatever else to write the code. So, that's I think what's happening across all of these systems. So, you're absolutely right that it starts with some kind of a UI that's more chat-focused, but it quickly goes beyond it, and you see it in M365, you see it in Dynamics 365, and you see it in GitHub.

Yeah, I think one of the other things, Kash, is that I think I said this in a previous earnings as well, which is the difference between a hoster and a hyperscaler is software. And the same is going to be true here. That GPT-4 example I gave is all software, right? The optimization even in the last year. So, we know how to use the software skills to take any piece of hardware and make it multiple X better. And so that's kind of where the yield will come. But as Amy said, while you're really going in building out the plant, you don't want to serialize it. You just want to go in parallel in all of these fronts and that's sort of what will compound over time.