Zelnick, Software Equity Research at Deutsche Bank. Truly delighted to be hosting this session with Microsoft. The format of the session will be a fireside chat. Got a number of topics to go over with Mr. Bill Duff, Executive Vice President, Chief Financial Officer. Not true. Of Microsoft's Global Commercial Sales and Marketing business. So Bill, really appreciate you coming here. It's great to have you. It's always good to see you, but even better here. Thanks, Brad. In Dana Point at the Deutsche Bank Technology Conference. Fantastic. Thanks, Brad. Appreciate the promotion. My pleasure. Anytime. You can have Amy or Satya; whoever it is, they can call me. If you just don't mind, to level set for everybody in the room that may not know exactly who you are, can you just take a minute to explain your role at Microsoft and what exactly your mandate is? Yeah. I lead finance for our commercial sales and marketing organization. You can think about it as the Microsoft commercial business and supporting all of our global sales organization and our marketing investments for the company. Awesome. You may remember that the last time we got together and had a proper sit-down was about a year ago up in Redmond; things are moving pretty fast since then. I would be curious to hear your perspective, perhaps to level set on a few things that are top of mind for customers today versus where they were a year ago. Yeah. It seems like forever ago, a year ago. A lot has changed, both in terms of the technology that is being delivered into the marketplace, but also how customers are really adopting this technology. I would say a year ago, there was a lot of interest, there was a lot of experimentation, there was a lot of how should I think about this, how should I think about how it would help my business overall?" but not necessarily clarity on how they were going to deploy, what the business outcomes were going to be. Over the past year, particularly as we got into the first half of this calendar year, it is really about getting more comfortable with technology. How do we scale? How do we govern it? This is really new technology; there is a lot of foundational infrastructure that needs to be in place in order to govern it. Then making sure we are driving the right business outcomes with this. It feels like it is getting to a more concrete discussion than we had a year ago; it is really people are starting to see, customers are starting to see the value, so they are looking at more opportunities more broadly across their operations. That makes sense. For us personally, it is reflected in the technology and how far it has come so quickly. Yeah. I mean, Copilot is a part of our workflows Yep Every single day in what we do. Yeah. We see it firsthand. But as customers go from pilots and where they were maybe a year ago into production, what do you see that distinguishes companies achieving scaled adoption and measurable returns from those that remain in the experimentation phase? Yeah, there's a few things. Certainly, I mentioned the foundational layer, so making sure that you have the right data structures in place. The models and agentic solutions are only as good as the data that they're reasoning against, so you need to make sure you have good data integrity. These models are reasoning against the right inputs overall. So foundational technologies, not only on the data side but also the governance side, the security side, making sure you have the right compliance in place. This is critically important for customers. As they think about deploying this technology, they need to have that foundational infrastructure in place. Then it's around, "How do I change my operations?" So really the thought process of thinking about I have employees plus agents. How do I redesign processes around thinking about leveraging agentic technologies in the overall process? A lot of times this requires taking a step back and actually looking at the fundamental operational aspects of the business and saying, "Hey, look, if we can use agentic technologies in a certain place, how would we restructure our operations around that?" That is a really, really important change management piece. Companies that are successful are really thinking about not necessarily just replacing existing ways of getting stuff done, but how do we re-architect processes. Then finally, you mentioned it a little bit in your question, which is how do I measure results? There is a real focus, and certainly from a finance perspective and the CFOs I talk to, they want to see measurable outcomes. This is not about new KPIs. This is not about creating a new KPI for how do I measure agentic performance. That is an important part of it, but it is really about business performance. Companies generally have business KPIs that they look at every day. It is not about redefining those. It is figuring out how this new technology can really help improve their performance on those KPIs. Those are pretty consistent as we talk to companies out there about this, deploying these types of new technologies. That makes sense, Bill. Everybody is trying to be on the front foot and achieve results and drive returns with AI. It is a top priority for customers. We get it. Budgets are not unlimited. Yeah Out there. As you talk with customers, where are they finding the dollars to fund AI initiatives? Has that conversation also evolved over the last year? Yeah. I would say the conversation has evolved, but it has gotten sharper. There is a lot of reprioritization, so as companies see the promise of this new technology, they want to make sure that they are investing for the future. A lot of the investment that is happening is not only on continuing to deploy this new technology, but making sure they have the right foundational architectures in place and the foundational infrastructure in place. We do see a reprioritization to invest in the future. When they look at their overall IT budgets, you can see this shifting towards AI. We are also seeing, though, incremental budgets being created. I would say that those are more focused on areas where there is clear return. You can see this probably most dramatically in developer productivity. When companies think about developer productivity, there is a very clear ROI. You see companies think about their budgets differently. They think about their development projects; they think about tokens as part of those or our GitHub Copilot as part of that overall budget. That really will be incremental budget. You also see it a lot in customer service, customer support. That is an area where there is very clear ROI. Sales efficiency, the area that I focus on, is another area of clear ROI. You do see incremental investments in those areas because you do see that clear return. Makes sense. I think the two areas of the business where we have seen the most visible success to date are within Azure and Microsoft 365 Copilot. Yep. I do not know if that is fair. Yep. You can tell me if you disagree with that. I want to hit on both. Starting with Azure, you disclosed a large amount of AI backlog is tied to Microsoft Frontier, but there's contribution from the broader customer base as well. Yep. Can you just help frame for us where enterprise customers are in the journey today and usage or any spending trends on Azure AI services? Yeah, we are seeing a real broadening of demand. The Frontier labs and our relationships with OpenAI and Anthropic are super important for us. They're big customers of ours, and they are really investing at the beginning of the cycle, so we've seen a lot of growth through both of those customers. We're seeing very broad demand. When we announced our results for last quarter, we saw an increase in our backlog of about $50 billion, a little bit more than $50 billion. That was all outside of OpenAI and Anthropic. So it's all non-Frontier labs, and that backlog is growing. So we're seeing a real broadening of demand. If you look at Foundry, which is probably our best proxy for this, you have over 100,000 customers who have doubled revenue Wow Year-over-year. So we are seeing a lot of companies across the spectrum. This isn't necessarily concentrated in a specific geo or a specific segment. We're seeing it very broadly. Of course, there's some customers that are driving pretty significant growth there, but we do see a broadening out of this, and we expect those trends to continue. That's what we're seeing in the market right now. Yeah. It's great to hear the dispersion and the broadening out in the different customer segments that you're serving. Unfortunately, it sounds like supply constraints will persist for the foreseeable future. That's what everybody's telling us, and that's not a Microsoft comment. Yep In isolation, obviously. Yep. You have to serve what's already in backlog first. Yeah Before anything else. As you get line of sight to additional capacity that's coming online, how do you internally think about prioritizing across various customer segments and deployment types? Yeah. For us, it starts with demand and understanding the demand out there and customer commitments. We are going to support our customers as they need incremental capacity, and where they are making commitments, we are committed to delivering that. But as you mentioned, there is a lot of demand out there. I think Amy talked about it in recent earnings. We expect that demand will continue to be greater than supply through the remainder of this calendar year. From a visibility standpoint, we are doing everything we can to increase capacity. We are driving efficiency with our internal R&D efforts. We are driving efficiency with our first-party services as well, to make sure that we are providing as much capacity as we can to our Azure customers. One of the things that we have talked about in the past is creating common architecture. So if you think about the customers using our either AI services or our infrastructure, our first-party services, our Copilots, and other stuff, they leverage the exact same infrastructure. We also support our R&D efforts on that exact same infrastructure. We have a very dynamic. We think about it very dynamically. We are looking at efficiencies across all of these areas, and we are making sure that we are doing the best we can and really optimizing to match supply to demand. It is a big effort within the company, and we are going to continue to focus on that as a critical growth driver for us. Because not only does it help with margins from a Microsoft perspective, but it helps us support our customers as their demands continue to grow. That makes sense, and I imagine it comes with some tough trade-offs from time to time. But I think especially in context of recent results, the year ended spectacularly, and especially as I think about now, let us go up the stack, and if we turn to Microsoft 365 Copilot, you reported impressive acceleration of new seats in the past couple of quarters. What do you attribute the inflection to? How much of it might be pricing versus packaging versus capability-driven on your side versus some kind of aha moment that customers are seeing in the market? I think the biggest driver there, if you sort of go back to H2 calendar last year, I talked about it as experimentation, but there is a lot of sort of limited deployments. People were testing out this technology. We saw this across a broad part of our customer base. They would deploy Microsoft 365 Copilot to a department, or they would do a limited experiment, and they were really trying to figure out a couple of things. One is their architecture ready for this? Do they have all the compliance and all of the foundational technologies in place in order to roll it out more broadly? What is the value there as well? What is usage going to be like? What is the feedback on the overall value of the solution? And I think what you've seen in certainly H1 of this calendar year, that acceleration you're talking about, is primarily driven by expansion. So these are customers who really started to look at it last year and really expanded this year. We're seeing a lot of deployments. We call them wall-to-wall deployments, where basically all information workers in a company receive a Copilot license, just like they receive an M365 license for their productivity software, just like they receive a Windows PC. You can think about it as making an information coworker productive. On the engagement side, what's interesting is the engagement of M365 Copilot is on par with what we see with Teams and Outlook, which are some of our most heavily used products. So this is what customers are seeing in their employee base. So they're seeing real value there. We think we have a differentiated position with Work IQ, which helps to the differentiated value. We continue to make significant improvements in the product. We roll out new features every time. I would say that the product today is significantly better than it was a year ago, and it'll be significantly better a year from now. We just see these improvements that we continue to push out into our customer base. So I think it's an element of all of these. It's becoming more of just a standard part of the technology stack you give information workers to be productive. Yeah. There's a lot of strong proof points of the value being realized. Yeah. But as things evolve, as customers roll out more compute-heavy agentic use cases with Copilot, you've introduced additional usage-based billing for Copilot Cowork and use cases leveraging the Work IQ API. Why is this hybrid approach the right one, and what are you doing to help customers forecast and get comfortable with the potential volatility of consumption-based pricing? I got to imagine that there's a customer that realizes they've burned through— Right Their commitment sooner than expected. How does that play out? It's a big issue. On the Microsoft 365 Copilot side, when we introduced it, for the most part, people are using it as chat. I'm going to do research, or I'm going to go into depth into something that's going to help me with a task, but it's very iterative process with a person. With the introduction of Copilot Cowork, this is about more of multi-turn, task-oriented agentic solutions. There's a lot more value and a lot more leverage. I could work with my copilot coworker to go accomplish certain tasks, and it would come back to me with that. These are longer-running agentic solutions, is kind of the way to think about it. There's a lot more value there. It's more expensive as well. One of the things that we're doing is we're making sure that we support a very high-value chat experience in Microsoft 365 Copilot. We think the consumption model is right for these longer-running agentic tasks that are really doing work on your behalf. We think that that's both a business model that works for our customers to more closely assign or closely align value with cost of the solution, but it's also something that works with our business model as well. Now, it introduces new complexities for our customers because before, one of the nice things about the Microsoft 365 business model is very predictable. I know exactly. I have an employee; I pay X per month for them to use our productivity stack, and it's very controllable and it's something that is very consistent and predictable. With Copilot Cowork, they're going to have to have a better understanding, so more visibility on what's being used, where it's being used, and give them the controls and the management to manage the overall cost. As part of Microsoft 365 E7 which was our new offering that we put out in Q4, was an admin center, Microsoft 365 admin center. We are providing the tools for companies to be able to manage this. I think this is a business model that will be broadly adopted, not only by Microsoft, but you can see others in the industry also adopting this. This is an area where Microsoft has a unique solution. We are very good at this. We have worked with companies for a very long time to how they control and manage their overall environment so they can look at this. We will give them the tools to look at this cost, make sure it is high ROI, and be able to manage it appropriately. That is our overall strategy. We think this is a great value for the customer, and we also think it is a good opportunity for growth for Microsoft. I think it is fair to say that Microsoft is a leader on behalf of the industry, right at the forefront in figuring out how customers are going to adopt and embrace and It is a critical one. Every CFO conversation I am in is how do I manage costs? How do I know I get value out of this? How am I going to manage this throughout my system? You talked about it earlier in terms of incremental costs and budgets and everything else. We are at the early stages of this, but I think, as you mentioned, we are at the forefront of it, and we feel great about our strategy, and we will continue to make progress. Maybe sticking at the applications layer, but turning to Dynamics. Last quarter, the company had called out some differences in customer behavior across different parts of the business. As AI becomes embedded in business processes, how do you think about the opportunity across the various Dynamics applications and workloads? Yeah. We are very excited about our Dynamics business broadly. I think we continue to see good strength in ERP. I think we mentioned in the last earnings, from a CRM cycle, we saw a little bit longer sales cycle, but overall, we do not see any change in underlying customer demand there. It is a super interesting area for us because we do think that agentic solutions will be extremely complementary and will be an important part of all of these solutions going forward. From a Dynamics perspective, making sure that we are thinking about this opportunity from really a systems-of-record type of an opportunity to a system of action is kind of how we think about it. You can think about how agents can reason against the data that companies have in these various applications and be able to really, I guess, agentify how they are improving their overall operations. It is a big opportunity for us. We think we are very well-positioned there. We have a very well-integrated stack across productivity software, so we think about it as part of one solution to our customers. We are excited about the opportunity. Cool. Maybe just switching topics a bit. Earlier this summer, Microsoft announced a $2.5 billion investment in the recently introduced Microsoft Frontier Company, embedding 6,000 engineers at customers, helping them deliver measurable return on their AI investments. How, if at all, does this differ from the FDE, forward deployed engineer, initiatives that many peers have announced, and why does AI success uniquely require this level of embedded engineering at the customer? Yeah. This was interesting. It's a big announcement for us, and it's something that our customers were really asking for. We've been engaged in a lot of customer conversation over the past couple of years, but obviously it's accelerated. How do we really use agentic software or deploy agentic software to improve our business outcomes? Generally it's around; I want to improve in a specific area, and we felt like we had the opportunity to work with our customers to develop these solutions to help them improve their overall operations. We're thinking about this as a business, so it's broader. It's not just us thinking about we have engineers that are focused on our technology and really maybe solving a tech issue with our customer. This is about working with our customers around delivering business outcomes. We think about it in terms of bringing in specific industry expertise. Obviously, technical expertise is a very important part of that. But this is us going into our customers or going with our customers and solving specific business issues. As I mentioned, it's a big business opportunity for us. We think about it as a business, and so it's broader than what we've seen out there as far as just FDE-type technology and FDE-type support. I guess as you explain that, I'm just trying to wrap my head around historically that which partners would do— Yeah Microsoft has the most vibrant partner ecosystem, I think- Right Of any tech company that I can even think of. How can you help partners develop these capabilities going forward so that they can amplify all of Microsoft's innovation and effort here and drive transformation for customers? Yeah. As you mentioned, we're very partner-centric, and we think this is a big opportunity for us to strengthen business with our partners through this, and really in two areas. One is partners will bring complementary skills, and so we expect to work with partners on helping our joint customers really redefine their business processes and deploy agentic solutions. There's going to be a lot of areas where our partners will bring in differentiated expertise, differentiated customer relationships and be able to help us, and we'll partner with them to deliver solutions to our customers. Overall, from a scale perspective, our partners have broad relationships with many different customers across many different segments. They will bring in their own IP; we will help train them in our technologies, but we'll have worked side by side on various customer solutions, and so we expect this to be a real amplification effort on Microsoft Frontier. So we think about Microsoft Frontier as not only a customer-centric investment for us, but we think partners are a critical part of us delivering that value. Makes sense. Bill, years ago, a software CEO, and I mean long ago, once said to me the best partnership he ever saw was a purchase order, which I always get a laugh at, but that kind of brings me to the next question. Right. When we think about the AI labs, they've been great customers and partners for Microsoft. Yeah. But at the same time, they're also increasingly looking to expand up the stack to the application layer, where Microsoft has an industry-leading position. You've always embraced co-opetition. Yeah where it makes sense. Right. Specifically, how do you see this playing out with the AI labs? Yeah. We really value our partnership with the AI labs, and we had an early relationship and investment with OpenAI, and I think both companies are where they are today, and we've both benefited from that relationship. We continue to think of ourselves as incredibly strong partners with the Microsoft Frontier Company, and we'll continue working with them. Obviously, they're very important customers of ours. Our strategy is really around a customer-centric strategy, thinking about how we can deliver value to our customers. We think a lot of customers will want a multimodal environment. We're thinking about ways for how do we provide the best solutions across our technology stack that really help our customers be successful. There are going to be areas where we are very complementary and partner very closely with the AI Frontier labs, and there are going to be areas where there might be areas of overlap competitively. But as long as we have our North Star and as long as we have our position with customers and doing the right thing for customers, I think we'll be very well-positioned. That makes perfect sense. Just relating back to the Microsoft Frontier Company and something Satya has focused on a lot is sovereign AI. Yeah. Just enabling access to AI and- Yeah Frontier models and whatnot, while also protecting customers' unique IP to use for their own benefit. Yeah. Can you just talk about the value of security and trust and how Microsoft is uniquely able to provide this in the AI era that we're in? Yeah. Sovereignty is an important part of our strategy overall. It's really a North Star for us when we think about it from a customer perspective. In the past, sometimes sovereignty was thought of in terms of data and data residency and things like that. But we think about sovereignty very broadly. It's the data and where the data resides and who owns the data, the customer's data. We think the customer will want to decide which models they use for which applications in which scenarios. The customer has their own unique IP and their own operational value. That's really how we think about sovereignty. It's about how do we make sure that we are allowing our customers, or we're enabling, not allowing, enabling our customers to compete most effectively, to retain their own IP, and really differentiate themselves in the market overall. It's a core part of Microsoft Frontier. When we think about our Microsoft Frontier solutions, that's how we think about it. But it's also more broadly how we think about our solutions in general and how we think about our strategy with our customers. You can see it in Microsoft 365 Copilot. It's multimodal. We'll allow customers to use the models that they want to use to help their business succeed. It's a critical part of our North Star, and it's something that I think we'll continue to lean in on. Perfect. Bill, one of the things that many of your management team and senior leaders we have met with at Microsoft over the years has impressed on us is the pressure to keep the feedback loop tight and to constantly evolve. Yeah. As you enter this new fiscal year, how are you evolving the strategy and goals within your organization? Yeah. From a sales organization standpoint, Microsoft Frontier Company is the biggest change that we are making from a go-to-market perspective. As I mentioned, this was really driven by our customers asking us to come in. If you think about how we are approaching our customers and the opportunity for us to help them transform with this technology is a really important change in terms of how we are working with customers overall across our overall commercial install base. That said, we take the same approach internally. If you think about Microsoft Frontier Company, it is not only that we are going to do Microsoft Frontier Company projects externally with our customers; we are going to do Microsoft Frontier Company projects internally. We have already done many of these projects, but we are doing more around sales operations, sales efficiency. We are taking a hard look at all of our processes internally to how can we run more efficiently, run more effectively, and it is really around sales leverage, sales effectiveness. We have a lot of projects internally that are going to be focused on how do we take our information workers, our sales force, our quota-carrying sales force and how do we make them more productive? We are going to agentify certain processes, and we are going to redefine certain processes around that. We have talked publicly about how we have done this in support. At a company our scale, we have saved hundreds of millions of USD a year in terms of support in just redefining processes, using technology in a certain way. It has allowed us to really re-architect our approach from a support perspective. We continue on that process, but we think there are many other areas that we can look at. Sales ops is a big one, and that's one we're going to focus on this year. As we look across the company, just as our customers are transforming, that's what we're doing internally. You've already kind of led into my next question, because I know Microsoft has always been on the forefront Yeah Of practicing what it preaches. Yeah. Specifically on leveraging AI internally, is there any way maybe you can give us more concrete examples or quantify the impact that this has had on the way that your team functions on a day-to-day basis and your ability to deliver operating leverage? Yeah. I would say even on my team, from a finance perspective, I have a global team that works with our sales leaders around the world. We have a bunch of different functions, including standard financial processes, such as closing our quarter, closing our month, helping provide business insights to our business partners. We have fundamentally changed how we have done that. If you think about our close process, where we basically look across all of our results and understand deviation, understand what is happening with the business, we are using agents and Copilot Cowork as a fundamental part of that process as we work through the close process. It gives us a couple of things. One, it helps us close much more quickly and more accurately, but it also frees up time for people on my team to really focus on what are the business insights, where is the judgment, what should we be focused on? How do we improve our business over time? It really changes the way you think about certain roles within the organization. Another part of my organization focuses on compliance, and so we enter in complex deals around the world. One of the ways that we had thought about this historically was you thought about high-risk deals. For whatever reason, we had a scoring methodology, and we would audit them, let us say, a couple, based on the high-risk nature of it. We are introducing agentic technology to fundamentally redefine that and say, "Hey, look, as we look at all of the deals we are doing around the world, what is our risk profile and how do we get insights out of that?" Those are just a couple of examples of just things that we do on a day-to-day basis, where we are both using agentic technology but also redefining the process and redefining what we do and where we add value as an organization. You can take that broadly across all of, we call it "MCAPS," Microsoft Customer and Partner Solutions. This is our broad sales force, and you can see it across all of our different areas, whether it is providing business insights, how is our business going, what do we need to do to improve overall performance, to a lot of operational aspects. When you get an organization as complex and as large as Microsoft, you can think about all of the different operational handoffs and the different teams that need to do things just to get a deal done and just to get a deal approval. We are looking at all of those processes and trying to redefine them in a world where you have agents and humans working together to become more productive and more efficient. The possibilities of how we are going to apply these technologies and where this is all going is just super exciting to me. It is super exciting, and it is early, I would say. Yep. We talked about Microsoft 365 Copilot and what we saw last year. It is sort of hard to imagine the difference in a year from where we were a year ago to where we are now, and we just introduced Copilot Cowork. We are just at the beginning of these journeys. As companies, we talked about Microsoft Frontier; companies are asking us to help them redefine certain business processes, drive certain business outcomes, but we are very early in that overall process, and I think we are going to learn a lot. The technology is improving dramatically, but we are at a fundamentally different stage than we were a year ago. Last year, experimentation. Now it is really around production, scale, governance. How do I drive business outcomes? How do I measure business outcomes? And you are starting to see it, which is really nice. Amazing. Well, with that, we are at the end of this session. Bill, really appreciate you being here. Is there anything you want to leave us with that I haven't asked you? Nope. This was great. Okay. Fantastic. Perfect. Always good to see you and even better here in Dana Point at the Deutsche Bank Technology Conference. Thank you so much, Brad. Thank you. All right.