Welcome everyone. I'm Brett Iversen, Vice President of Investor Relations. This is the 8th in our series of quarterly videos focusing on strategic areas that are top of mind for our investors. Today's discussion will cover Data and AI, including our view of the growth opportunity, our differentiated offerings, and how customers are already using our solutions today.
We've brought together two of our key leaders to answer your most frequently asked questions. We have Jessica Hawk, who leads Data and AI Product Marketing, and Eric Boyd, who leads Azure AI product development efforts. As always, we'd welcome any feedback on specific topics, the format, or other suggestions you might have after you view the video. Please reach out to our Investor Relations team directly with any feedback. With that, let's kick things off. Thank you both for being here.
Jessica, maybe you can lead us out with a little bit of framing on, you know, how do we talk about data and AI together. How would you help our audience think of those two concepts together, and why we think that's important?
Sure. You know, I think you've probably heard me say data truly is the lifeblood of AI. AI is without question having just an amazing moment right now in the marketplace. It's truly a call back to how is your data organized? It is a complete get your data house in order moment as customers get really excited about what these AI capabilities can do for them.
You know, there are some just, you know, horror stats out there about data growth, data sprawl, data leaks everywhere. The most recent one I've seen is that we're projecting 181 zettabytes of data by 2025, which is a ginormous amount of data swirling around the ecosystem.
For sure, just in terms of getting ready and organized to take advantage of the moment, it is a complete call to action to accelerate your digital transformation journey. You know, last year we announced the Microsoft Intelligent Data Platform in anticipation of these opportunities that we're bringing because we know that customers need to get the data organized to take advantage.
Think of it as the way we describe the full breadth of the data offerings in Azure that organize around operational databases, which are the data state systems that have been supporting apps for decades.
Right.
Analytics capability to make sense of all the data without impacting the performance of the applications that are running on the operational data stores, and then the governance that's required, especially with the, with the ginormous growth of data in our customers' environments to ensure that the right people are seeing the right data at the right time. You know, we organize the business at Microsoft as a data and AI unit because truly these two things have to go hand in hand.
Yeah. Our phones have been ringing nonstop with some of the announcements, lately as well. Eric, I'll bring you in as well. Also thank you both for spending the time today.
Of course.
One of the questions we ask in all of these sessions, given our financial investor audience, how would you think about the growth opportunity? How would you help frame that for folks watching?
Yeah, I mean, you know, Jessica sort of alluded to it, you know, with the technical term ginormous. Honestly, when we think about it, this is really the most significant transformational technology breakthrough in our lifetimes, which this, that's saying something, right?
Yeah.
I mean, we've lived through mobile, we've lived through the internet. Like the world has changed dramatically, and we really feel like we're poised to see that happen again with this transformation in AI. you know, we really feel like Microsoft is super well positioned for this as well. I mean, we have the foundational infrastructure, everything from Azure to our developer tools with Azure Machine Learning and our Cognitive Services, all the way up to our SaaS offerings all across the company from Office to GitHub to, you know, everything that we offer.
Really positions us in a really very strong way. We're really excited about the opportunities that we have here. We also see these as additive opportunities as well. These are new things that people are coming to us for.
If you go and tell a company, "I can make your developers 55% more productive by using GitHub Copilot," that's something they want to go after. You know, that it saves them, it saves them money, drives their productivity. These are really additive scenarios to the, to the work that Microsoft already does across the board.
Yeah. ginormous works in their Excel models as well.
It's yeah.
They're excited about it. Jessica, back to you. In terms of data and AI and its relevance with the broader Microsoft Cloud vision, you know, how would you help people think of all of that together and where we're differentiated?
Sure. Well, in the halls you'll often hear me saying it's really an all boats rise moment.
Yeah.
What I mean by that is, if you think about the Microsoft Cloud, we have our first party offerings, so things like Teams and Outlook and GitHub. The neat thing about what we're doing is we are taking all of the work that Eric's team is engineering, and we're infusing it into those first party products from the get-go. What does that mean? It means that our testing cycles, our engagement loops between our engineering teams, our internal engineering teams are our largest and probably most challenging customer-
That's right.
when you think about it. What an incredible opportunity from a software development lifecycle perspective, because Eric’s team is getting this continuous cycle of improvement and feedback and scale.
Right.
The scale that these platforms operate at are, you know, second to none. We take that same incredible throughput and feedback loop, and then we put that into our third-party offerings, so all boats rise. Our customers are getting tremendous AI-driven capabilities through our first-party products. That same technology is now available in Azure as a third-party service. What we're seeing is imaginations are starting to open up now.
People are starting to really understand This is unique. This is the biggest innovation in our lifetime. They're seeing it and experiencing it and benefiting from it in their first-party solutions that they're using from Microsoft, and now they can go build their own internal or customer-facing applications using our Azure services. It's truly an all boats rise moment for our customers.
Yeah. I have to think it's super comforting for our customers to hear we're using all of the same tech that we're wanting them to embrace as well. Back to you, Eric, on the, you know, I think Jessica mentioned Azure AI Platform. Maybe you could give us a few more details to help people that aren't as close to it as the two of you.
Yeah, no, absolutely. you know, for us it has always started with research, and so we've been benefited by having Microsoft Research as a part of Microsoft for something like the last 30 years. We've been first to human parity in a whole number of fields in AI, and what we then work to do is how do we bring that work that's happened in the lab and, makes a lovely research paper, we need to now make that work for our enterprises. It needs to be at scale for all of the things that our customers need to do. Jessica alluded to, you know, the benefit that we have.
You know, we're really privileged as a company to have so many different businesses that we can partner with and learn how is this stuff really working, and how is it impacting businesses in a positive way? We then bring that into our products and make that available through Azure AI. You know, all of this is built on the strong foundation of Azure.
You know, being able to have access to thousands of GPUs and computers and being able to pull them together. We then provide developer tools. Azure Machine Learning is a layer that developers can build and train and manage the full operational life cycle MLOps of their models. At the next layer up, we provide a suite of Cognitive Services, and so in areas like vision, in language, in, you know, speech recognition.
Right.
Of course, our partnership with OpenAI shows up there. These are models that we've built that we make available that customers can just go and call directly on their own. At the top level, we have applied AI services, where we look at the most common patterns that we see on these Cognitive Services, and then we package them.
Form Recognizer makes it really easy to take this invoice and get it into that database over there by using our vision services, our optical character recognition and things like that. Putting that whole package together for developers. We also use all of the breadth that we have across Microsoft, and so it'll show up in developer platform where we'll make, you know, AI available.
It'll show up in the Power Platform, where all these AI models are available, and it shows up in Office. All throughout our suite of products, we're bringing AI to our customers, leveraging this foundation that we have on Azure AI.
Yeah, I think that breadth of offering, and I'll probably come back to that in a little bit.
Yeah.
Again, another area that's really kind of excited, maybe even caught our investors by surprise at how quickly we have so many different things out in market versus conceptually years out. One follow-up to one of the things you mentioned. OpenAI partnership, again, our investor relations team would get tons of questions on this.
Yeah.
Some of which we answer, some of which we don't. In terms of the importance of it and how we think about it, what can you help share there?
I mean, it's been a really transformational partnership for us as a company. you know, we see these large language models as being foundational to driving this transformational change that we see really happening across the industry.
We realized early on that OpenAI had the same vision that we did for how you could continue to grow and scale and make these models way more powerful, by continuing to train, and so that really forms the bedrock of our partnership. From there, we just see tremendous advantage. I mean, we've renewed our partnership. We announced, multiple years and multiple $ billion, really to accelerate this work going forward.
You know, everything from the work that we do on supercomputers, we learn tremendous amounts about how to build the right infrastructure, which we then make available to other customers through our Azure platform, as well as make these services available, and we really get a first mover advantage by being able to bring these, you know, to customers all across our industries as well as throughout our internal products.
Tremendous momentum there. Of course, as a part of that, you know, we're the exclusive cloud partner for OpenAI, just really strong benefits for both companies in this partnership, and we're really excited to see where we go with that going forward.
Yeah.
If I could just add on to that for a second.
Yeah, please.
You know, I mentioned in the beginning when we were talking about what happened in the, in late fall, early winter, there's this interesting moment. We have this in other examples, mobile would be one that I would mention as well, where technology breaks through into the consumer mindset.
Yeah.
You know, Eric and I's children go to school together, and I would argue they might be some of the most popular kids in that middle school right now because their parents are somewhat associated with that amazing thing that was launched on the internet last year. When I talk about customer imagination getting sparked...
Yeah
I think it used to be that technology tended to follow what consumer demand was creating. An example, pick your favorite app. It used to be when companies would write business requirement documents to go build something, if, you know, back in the day, it was you would build everything you were gonna do for the web presence, then maybe you'd get to a mobile.
It was almost always like the last part of the Word document. Then consumer apps started to come out, and demand, user experience demand went from just out in the consumer experience to, "I want the same amazing capability and delightful experience in my, what I do at my office too.
Sure.
With the introduction of ChatGPT out in a consumer-facing way, that's what I mean by imagination getting sparked. I think that the bar for excellence is much higher, and there's gonna be less patience with waiting for those types of delightful experiences.
What we're seeing is our customers are, you know, maybe unlike with cloud, when cloud first came out and there was some cloud adversity or concern, we're in the opposite position, where I think people recognize there's huge value if you can create those same delightful customer experiences with your commercial applications that people are getting accustomed to on the consumer side. We're seeing it already. There's articles and news stories being written.
My favorite one is from a CNN article where they were citing that, you know, it's a real estate real estate agency, and the agents are saying like, "I will not do my job now without ChatGPT." That kind of speed, when you think about how fast that happened, they went out in December and that news article was written in January. I think we're gonna just see, like, this massive leaning in and desire to innovate together, which is for technology, it's a wonderful opportunity for all of us.
Yeah. The energy from customers has been-
Yeah.
honestly like nothing I've seen any-
Yeah.
on any product anywhere.
Across the board, right? Yeah.
It's just really amazing.
Yeah. It's on that, I know that we're in the right space when I'm not just getting questions from, you know, coworkers or friends that are also in tech companies and other places, but, you know, my girls are coming home from school.
Yeah.
Asking me questions about it and talking about how they're using it with their friends. My mom's asking questions about it.
Exactly.
I mean, you can see.
The full span.
Totally. Like, everyone has very specific case studies that are valuable to them that really resonate, which is a great spot to be in. On that, it was a good segue then, Jessica, on, you know, we've had as a, as a very senior marketing leader, you must have been excited about the steady drumbeat we've continued to have in terms of rolling out services. You know, what would you share? I mean, It's hard to pick favorites, but what are some of the things that have been resonating?
Sure.
that you'd share with our audience?
Well, I think one of my personal favorites is GitHub Copilot, and it's not just because I used to write software, but really. It's because, I think we should be very grounded in the experience or the journey that we're all going through together, which is this is a true technological advancement. I think folks are waiting a bit to see how it plays out in terms of what's going to happen with the developer experience, for example.
When we announced GitHub Copilot at Ignite, you know, we were really excited about the capability. You know, as a person who used to debug software all the time, I am not interested in doing that. I never was. It was exciting to see some of that automation. I think people had reasonable questions about where this is all headed. You know, fast-forward six months, and it's pretty exciting when the GitHub Copilot productivity stats, you know, Eric mentioned it, 55% more productive. That's incredible.
Right.
The one that really gets me is the excitement that the developers feel over their job satisfaction. I think it's something over 74%. They feel more than 70% more empowered to do the work that really matters, and that's what we're going for.
This is truly technology that's gonna innovate for productivity's sake. When you have people who are more excited to get to their desk every day to do their jobs as a result of some technology innovation, like, that's where we're all benefiting. That one jumps out at me. Viva Sales is pretty exciting.
Yeah.
I used to run a sales team. Certainly it is similar to debugging in the software world. If you're a seller, CRM updates are pretty much the very last thing that you ever wanna do. Yet, you know, sales leaders and marketing manage, everyone's always, you know, "Get your CRM updates in there."
Just seeing that some of that sort of repetitive work, again, same concept, like take the part out of the work that people truly don't enjoy and give them more time to do what they do enjoy. Viva Sales is another one that really jumps out at me as pretty exciting, and I think it's a just like increasing developer capacity, which is a concern for pretty much every customer that we talk to.
Right. Right.
Hard to find talent.
I think sales managers and sellers are feeling the same type of pressure. I think we're gonna see a similar cycle of adoption with Viva Sales.
Yeah. I was in the U.S. subsidiary of Microsoft a few jobs ago. My sales friends would definitely agree with you that CRM updates is not.
Yeah.
What they would consider the most valuable portion of their day. If we can help simplify that. You know, Eric, similar to what we do with the platform, maybe, and this is to both of you, maybe a few more details on Azure OpenAI services. You know, Jessica, you could help breathe life into it, you know, from a customer perspective with some examples and things that we're seeing people do.
Yeah, sure. The Azure OpenAI service really is to provide our customers with access to all of the latest models that, you know, we've partnered with OpenAI to create. obviously GPT-3.5 was sort of... Three and 3.5 were sort of the first versions that people could sort of get their hands on to experience the power of these large language models, and then Codex, which powers the Copilot experience.
DALL·E 2, which is a visual image generation one, which has seen a tremendous amount of uptake. of course, the amazing amount of interest we've seen around ChatGPT, and so being able to get direct access to that model, really for use within their customers and for within all the different things they wanna do for their applications.
You know, we've worked at Microsoft really hard at sort of developing that through all of our different applications as well. I mean, Jessica touched on that really powerful cycle that we have of being able to work with our first-party customers and really learn where is this working and how do we need to change the way these offerings are set up, and how do we need to, you know, really tune these models to work best.
You know, through that, we're able to offer just much better services on the other end to our third-party customers, having proven them at scale, having proven out the use cases with them, and really made them work great.
Yeah. I'd say, you know, in terms of customer adoption, you know, we have over 1,000 customers who are using the service, and if I were to try to pick a couple that jump out at me, CarMax is a great example that I love to talk about, especially again, because it's a story of truly empowering an editorial team, where they took all of their user reviews. So, you know, if you think about the last time you tried to go pick a restaurant for dinner.
Yeah.
The stars aren't really getting it done anymore. People are reading through, or at least I am.
Yeah.
-to understand more about this restaurant. CarMax has a similar situation when you think about the service they're offering. It's, you know, I'm going to buy a car, and I wanna go look at the user reviews on the car beyond the specifications that they had already coded into the platform. User reviews are becoming a very good signal for buyers. But they have a small editorial team, and the last thing that editorial team wants to do is read through user reviews to try to summarize what those reviews are saying.
Because they had already gone through the digital transformation process of getting their data up and running into Azure, they were able to apply the Azure OpenAI Service, which their senior development lead described as a comically easy thing to do, and I'll get back to that in a second, to then go learn from the sentiment in the user reviews, and you might say,
"Well, we've had that for a while," but then go generate findings from that so that the editorial team could focus on creating the most engaging experience for somebody coming to try to learn about the cars. They estimated that it would have taken them 11 years to do what they did in just a few days and weeks.
That's amazing.
using the OpenAI service. The team is, you know, doing their best work. CarMax is one example. There's a global Austrian construction company called STRABAG. Very similar situation. They were already getting the need for digital transformation, so they had taken their data estate and done the work of centralizing it up into Azure.
Similarly, they were able to take the Azure OpenAI service and use it. They trained it on just a few months' worth of data, which, you know, that's a small sample. They went and looked at risk for their projects and were able to score Azure OpenAI's ability to go find that risk, and it was effective in 80% of the cases. If you're thinking about companies looking at risk projects, are they on time? Are they on budget?
Are there considerations about what's happening at the job site that might put this project at the top of the list that's worthy of a phone call to see how things are going? They have the ability now to get in the driver's seat, not find out after the problem has occurred, but actually start to seek signal earlier. It's because they're using a service that can run across so much data.
That is beyond our ability to keep a developer coding for all the possible different inputs. That's what we mean when we talk about generative AI, that it's able to learn and create new experiences that... which it's just impossible to keep up with from a software developer perspective.
They're in a position now where they can go find that risk ahead of time, which is job number one for companies like that. Then I said I would come back to it. I think this is a rallying cry for our customers who have maybe been looking at digital transformation and getting the data house in order. You know, the other part of my portfolio, I've got all the databases and-
Yeah.
You know, customers are trying to make it a priority. There's no question that there's massive capability and advantage that comes from getting your data house in order. Sometimes it's hard to find justification, particularly in accounts where the system is there, and it's been working.
Right.
Why now? I would say that this moment as again, people are starting to internalize what it can do from a true business impact level, I mean, the web experience with ChatGPT is fun. Nobody's knocking it. I've used it a ton. I had it help me with some naming things recently.
These scenarios I just described with CarMax and STRABAG, they're real, true, hard-hitting problems in the enterprise that they're now saying, "Okay, now I have a real business case for doing that migration." We've been saying migrate to innovate. I think we're seeing a different level of conversation with our customers right now about their excitement to accelerate their data transformations.
Yeah. I love all these productivity themes that both of you have mentioned. Even the creativity side, too. One of our analysts called me soon after they were able to start playing with some of the early models, and it was a visualization. He was like, "Brett, this isn't just finding a visual that is closest to what I'm describing. This is creating and authoring brand new content." We were supposed to have an hour-long call on a whole bunch of topics, and we spent it all on that one topic because he was so excited.
I don't think, you know, given risk and this topic is so big, it's always important as we're talking about AI and kind of what we're doing now on our path forward, to touch on the responsible AI portion as well, because I know people want to hear how we think about that, et cetera.
Sure.
Maybe you can help us with that piece as well.
Well, I'd start with what you just said there, which is there's an education journey...
Yeah.
That we're all on, right? Why is this moment with these large language models, we're using the term generative now. What does that mean? How is it different from what we had with, you know, neural network capabilities back in 2017? I do think this is an opportunity for us to live our mission, where we're truly empowering our customers.
There's a lot of education to be doing. We've been working very diligently on building out responsible use frameworks as an organization for many years now. Starting in 2017, we started to invest in a focused part of our both our legal team and our engineering teams to build out our principles. We know it's a huge issue, you have to start with principles.
We developed our six principles, and then our engineering team has basically been customer zero for how you put those principles into practice. What's really interesting and where I get excited in terms of trying to help our customers because I think there is a desire to jump into the pool, there's also a recognition that this needs to be done very carefully. As with any technology, once it comes into an organization, despite the best of intentions.
Right.
There's plenty of CIOs who could validate this, I'm sure, things tend to take a life of their own. I think similar to how customers think about data governance, they're thinking about responsible AI governance.
We did something that I think is somewhat unusual for us. We took all of the learning that we developed between our engineering and our legal team that's really focused on helping think through these issues, and we made that experience and the standard with which we operate against those principles available to the market.
Last June, we put the responsible AI standard out for customers to learn from, to adopt. We always say this is a journey we're all on together. We're trying to help recognize that the principles are great. Customers wanna say, "Okay, but when I get fingers to keyboard, how do I think about doing this?
Right.
Putting that standard, think of it as a playbook, out there for folks to learn from. Of course, we continue to just infuse AI capability through tools and into our products directly. Our company customer leaders, who I think need to meet the demand and the excitement from the C-suite, from the board, from investors with an understanding that they have an ability to go execute against the opportunity and do it in a responsible way. We're constantly infusing that directly into the products, and have continued to put out a bunch of dashboards and tool sets that customers can use to just go inspect-
how are we doing against the principles? Is the model being used the way it needs to be used? Is there transparency in how the technology is creating insights or generating new content? Transparency tends to be particularly important in this space.
Right.
Just to give them a way to go and seek signal and make sure that things are being used the way that they would want to be.
Got it. Given the size of the opportunity and the excitement around that, you know, in the finance world at least is, well, how much is it gonna cost?
That's right.
You know, like, specifically kind of the capital spend over time that we think will be needed to scale this vision the way that we think it will scale. You know, no math specific needed, Eric. How would you help us, if I could adopt you into the finance group for a minute to help us think about this, what would you share with our audience?
I was a math major, so I can do some of that.
I have no doubt.
I mean, obviously, these models are very large and use a lot of data, so there's a lot of compute used in them. We certainly see that as a trend that we will see. That said, we also we use these, you know, the compute fabric that we have are all of our GPUs, we use for all of the different use cases we have all across the company. By having sort of this common fabric, we really get economies of scale.
Right.
We're able to use, you know, the same sets of GPUs for Bing as we use for Office, as we use for, you know, a lot of our internal training and all of those workloads really sharing that common fabric. We see a lot of benefits coming from that. Additionally, you know, these models are gonna get cheaper over time. We're already seeing that, I mean, from sort of the first versions that we get of a model to sort of where we get them in production. We're able to dramatically reduce.
Right.
The cost that it takes to serve. You know, we do a lot of different work on it, everything from, you know, looking at, you know, the individual kernels of code that are running on each individual piece of hardware and optimizing that, to thinking about how I can use multiple GPUs and machines to sort of pipeline and stage things together and all. A lot of optimization techniques. You know, with so much focus and interest on this, we're just gonna see more.
Yeah.
We're gonna see more optimization really coming around that's really gonna drive this benefit to it. I think we'll see a lot of benefit from that. The other interesting thing to think about as you think about sort of this, you know, the cost associated with it, you know, AI has gravity to it being able to really, you know, have these models, and then they need to run on data, the closer you can sort of bring that data to the model, you know, the speed of light is a factor.
Right.
We can't go faster. How close can you get things really brings benefit to customers. If you're looking for the fastest way to sort of serve these models, the fastest inferencing speed and all those things, bringing everything together is gonna be a big benefit. Again, we see a lot of potential with that in the work that we're doing.
Yeah. I'm sure it's only additive to customer, you know, conversations we're already having and just one more compelling reason.
Absolutely.
Yeah.
I mean, Jessica sort of alluded to it, right? Like customers that are on the fence of, should I get my data state in order? Should I figure out, you know, how do I do my digital transformation?
Right.
Is it finally time? You know, there's so much interest, everyone is sort of moving in this direction. You know, the people who are the laggards are starting to worry about, "Boy, we might become the laggards." How do they really jump on this wave and really accelerate the work that they've been doing?
You know, before I close this out, one last question maybe for both of you. Yeah, just your final thoughts, you know, and/or kind of what do you think about in terms of what's next? Like, what, where would you leave our audience in terms of closing thoughts on this topic?
Yeah, I mean, I opened with like, you know, I think this is, you know, one of the most significant technological transformations in our lifetimes. The other exciting part with that is we're literally at the starting line.
I mean, we've figured out some of the things that we can do, but we're constantly learning and iterating with our internal teams, with our customers, and sort of starting to see where all of this can go. I mean, Jessica touched on the use cases. They're like all over the map in terms of the things customers are doing in every single industry that we see. You know, there's just so much more to come in this.
you know, we haven't reached diminishing returns in any of the things we're starting to see as we continue to build these models and make them larger and really prove them out. So there's just so much opportunity just right in front of us that, you know, I think these next few years are gonna be really exciting and fun to just sort of see where they go.
Yeah.
Yeah. I would just add, I was on the phone with one of our large, GSI partners this morning. I would say it's a where do you wanna go with your imagination moment, and it's been a real, rallying cry for those partners who are taking those customer calls just like we are around get that the opportunity is here. It's amazing that the technology is here. I can't believe how quickly you guys pulled all this together. It's available for us to go build on now too. Help us think through where do we get started.
Right. Right.
There's a lot of, you know, ideation workshops going on, and I love that it's a. You know, I've been around technology for a couple decades now. It's, it's a, it's a total business and IT together moment, which is, you know, I think personally pretty satisfying for folks that work in this space. Lots of conversation around, we see the opportunity use cases are really quite broad.
Let's get excited and focused on where do we wanna start, and then let's just go. A lot of our partners are getting super engaged in the conversation in terms of helping customers continue with their digital transformation and then prioritize where do we start with all the opportunity. I think, you know, us continuing to support our partners as they build out those practices is pretty exciting. I would say for sure it's again supporting customers on their journey with regards to responsible AI.
Yeah.
You'll see lots coming around, more support that we can provide our customers directly on that. Finally, we're gonna continue to infuse all of Eric's team's goodness into the rest of our first-party offerings, so you can be watching for that as well.
Yeah. When the space is as big as it is, it totally makes sense to me that partners are like, "Hey, we're really excited, but help us know-
Yeah.
where to start," 'cause there's so many places you could start. Well, I know you both are incredibly busy given how much this space is on fire. Thank you for both your time today as well as your insights. I appreciate it. We thank everyone for watching.
We're excited, as you've heard, about the data and AI opportunity ahead, our position in this market where we feel like we're in a leadership moment, which is really exciting, and the value customers are already seeing with our solutions. We're glad we had some time to talk to you about it today. Thank you.