# Satya Nadella, CEO of Microsoft, on the future of enterprise models with Applied Compute.

*Applied Compute* · https://www.youtube.com/watch?v=g_iUdhxpc4k

## Summary

Satya Nadella, CEO of Microsoft, sits down with Applied Compute to discuss his article "A Frontier Without an Ecosystem Is Not Stable" and the future of enterprise AI. He argues that firms are learning systems and should build their own token capital — evals, context layers, and fine-tuned models — rather than locking into a few frontier model labs. The conversation covers Microsoft's Work IQ and Copilot approach, the enduring role of human capital alongside token capital, compute supply constraints, refounding Microsoft for the AI paradigm shift, and why software and extensible, model-diverse platforms remain durable.

## Key points

- Satya Nadella argues there should be as many models in the world as firms, because a firm is a learning system compounding tacit knowledge, and AI should extend that compounding.
- A stable platform requires more value to be created around it than by it; if all value accrues to a few frontier model labs, the economy and political economy won't tolerate it.
- The vast amount of valuable data sits inside enterprises, so the unlock is firms hill climbing with their own R&E, context, and traces, possibly fine-tuning their own cost-efficient models.
- Companies should be outcome-oriented: define the eval that matters, welcome all models, avoid lock-in, and control their own token economics.
- Microsoft's Work IQ is the context layer from Microsoft 365 — communications, projects, calendars, documents — that Copilot mixes with any model; an HR onboarding example shows token-efficient agentic systems.
- Human capital is essential: 'cognitive coverage' over machine output is a valued job, and there is no chance of creating token capital without human capital.
- The hard part of a technology transition is the business model shift, and software has shifted from pure input to output; Microsoft must change how it works across every function.
- Software still has value — Excel artifacts, Teams interfaces, agent sidecars — but pricing must adjust, and no one can sit on what they produced.
- Microsoft builds every layer of the stack to be ecosystem extensible and model diverse, because positive-sum interoperability grows the market and their share.

## Notes

### A Frontier Without an Ecosystem Is Not Stable
- Satya Nadella's article stems from two questions: what is the future of a firm in the age of AI, and what is the nature of AI platforms and ecosystems.
- AI is a learning system, unlike traditional digital tools. A firm is itself a learning system built on human capital and tacit knowledge, so in the age of AI there should be "as many models in the world as firms in the world."
- If a finite set of frontier models has learned everything differentiated in the economy, the system collapses; a stable economy needs broad-based growth, not value accruing to one or two labs.

### Owning vs Renting Intelligence
- The interviewer cites four tailwinds: the "token apocalypse" of cost, data protection and model availability risks, labs moving up into the application layer, and open source models getting very good.
- Nadella's framework: start with the outcome or eval that matters for a business process (e.g., supply chain planning), welcome all models, then use your own context, data, and traces to fine-tune a cost-efficient open-weight model that hits the same eval.
- This keeps you grounded on results, avoids lock-in, and puts you in control of your token economics. "You're locked into one thing, which is the monotonic improvement of your outputs."

### Work IQ and Learning Loops
- Microsoft 365 holds the unstructured database of an enterprise: communications, projects, timelines, calendars, documents — the tacit knowledge and pointers to business processes.
- "Work IQ" is the context layer mixed with any model; Copilot was built this way. An HR onboarding agent example shows token efficiency: context as input tokens, avoiding regenerating output tokens on frontier models.
- Then firms can hill climb: evals, context, models, and use trace data to fine-tune cheaper models continuously.

### Human Capital and Token Capital
- Nadella compares AI to the 1980s shift to knowledge work: nobody predicted 3 billion typists, but we invented knowledge work. Now we need a reinvention.
- "Cognitive coverage" — humans interrogating machine output, like test coverage — is a job that will be valued, taught, and wage-supported.
- Even with RSI, the only thing valuable is the human capital controlling the AI; there is no chance of creating token capital without human capital.

### Refounding, Software, and the Platform
- Technology transitions are one-way doors; the harder part is the business model shift. Software has gone from pure input to output, changing Microsoft's production function.
- Software isn't dead: Excel artifacts, agent sidecars, and new middle tiers (e.g., so AI reasoning loops don't crowd out human interaction) mean the need for software will explode, though pricing must adjust.
- Microsoft builds every layer of the stack to be ecosystem extensible and model diverse, because positive-sum interoperability grows the market and its share.

## Transcript

[00:00] Well, Satya, thank you for joining me and thank you for hosting me at your beautiful offices here at Microsoft HQ. I'm super grateful to be here and I'm looking forward to talking to you about what the future of AI in the enterprise actually looks like. Our mission is to help companies own their intelligence by being able to train their own models and improve those the more they use them. So, when you recently wrote this article that got a lot of traction on X, I think it was titled "A Frontier Without an Ecosystem Is Not Stable", it really resonated with a lot of the things that we think about at at AC. And, you know, I wanted to maybe start the conversation talking a little bit about uh you know, your core arguments

[00:46] in in that piece, which is you know, the stable equilibrium is really companies building on top of frontier capabilities rather than being attached to one or two frontier model labs. I was curious if you could maybe talk a little bit about what that looks like in practice, um how Microsoft is approaching this ecosystem play, and what are the maybe the practical things that are being built on the infrastructure side to make it happen. >> Yeah, no, first of all, yes, it's great to have you here on the Microsoft campus and also it's great to be partnered with you as you're building out um AC and uh we're building out essentially this frontier ecosystem. >> Um to me, the uh the motivation for both that article and what we're

[01:32] doing comes stems from understanding, I would say, two uh things or answering two questions, perhaps. One is what's the future of a firm in the ecosystem or in the economy uh in the age of AI, as well as what's the nature of the ecosystem and platforms uh with AI, right? So, let's start with the first one which which is AI is a learning system, right? So, it's sort of a not a traditional digital technology or it's not a technology like all the other digital tools we built. And so, therefore, one has to sort of think about in a world where a model can learn anything. >> Yeah. >> What are the limits to that? Or in other words, how should the firm itself think of itself going forward? And so, my simple thing is there should be as

[02:34] many models in the world as firms in the world, right? Because after all, what is a firm? A firm is a learning system that today is mostly about human capital with digital tools every day compounding the tacit knowledge that the firm has and how it produces a good or a service that's valuable in the economy. >> Totally. >> Right. So, if that is the definition of a firm, my question is then in the age of AI, it should compound because of both its ability to create the token capital that then is being used in conjunction with the human capital and then the compounding loop between the two, right? So, that's sort of one part of it. In order for that to be true, then you do need to have a platform that

[03:22] is really enabling every firm in the economy to be able to create its own token capital. It can't be, "Hey, look, I have two frontier models or three frontier models or whatever, some finite set that have learned everything that is differentiated today in the economy." Because then it collapses, right? So, there is no stability to even the economy and definitely the political economy won't tolerate it because you kind of need a world that is, you know, there may be very successful companies. This is not about sort of any one company's success. It is about an economy being stable because there is a lot of economic growth that is more broad-spread. And so that's sort of the the two part and that's why I wrote that.

[04:10] And you know, and then we can unpack that. I mean, in terms of each of the layers. >> Totally. Yeah. And and I guess like maybe drilling in on that point about the stability of the economy, what do you think just breaks down when you have all the value accruing to maybe a small set of model companies? >> Yeah, I mean, if you think about it like definition of these tech platforms at least that I subscribe to is more value has to be created about the platform for the platform to be stable. Because >> Yeah, it's like the definition of >> Yeah, it's otherwise it's kind of like, you know, if you you need customers in order to be successful. And so that means customers have to be successful. And that is a little bit of

[04:49] the the simple logic on it. >> Yeah. >> But more than that, right? I kind of look at this and say today, first of all, I think we've got some very cool attributes to these models and we now know that what fuels that is of course compute, some great talent, but also a lot of data. >> Yes. >> And so I sort of look at it and say, hey, we have now created a lot of value in these models using let's call it public data and great talent and lots of compute. >> Yeah. >> But where is the vast amount of data still? It's all in the enterprises or in the companies. And so where the unlock at least if you sort of it's not it's it's not a zero-sum, right? It's about what is the positive sum way for, let's say, to

[05:38] achieve what I think has to be the goal here collectively. Can the GDP inflect to 10%, 15% or what have you? >> Yeah. >> And I think that will come when every firm is able to take even a frontier model or an open weight model or whatever and then hill climb using their R and E, right? Which is the company itself. >> Yeah. >> And in that process even build maybe even their own model or two which is much more fine-tuned to the their needs and doesn't have to spend all the tokens >> Totally. >> on a frontier model, right? So that I think is where the direction of travel is. >> That makes makes total sense and it's a it's a perfect lead into what I was going to ask next which is how do you make this decision about when

[06:24] to own versus rent? Because I think, you know, you've said this before people will use other people's AI, they will build their own AI, but when do you decide to do what? And from our experience at AC, we've like seen kind of like four, you know, four sort of different tailwinds that are kind of driving this all kind of recent actually. You know, one is just the token apocalypse, things are getting really expensive. Two is, you know, with maybe the most recent Claude Fable release and the stipulations around it, data protections, government stepping in, maybe the models on the frontier APIs won't always be there. The third kind of being like labs are moving more into the application layer and people are like,

[07:05] oh, I want to differentiate on multiple levels than just depend on frontier models. And the fourth being open source models are starting to get really, really good. >> Yeah. >> I'm curious and this data point was a really, really good point. What are the are there any other major reasons why you think companies might choose to own their intelligence? >> Yeah, I mean, I I think you you you hit on the most of the criteria, but the way I I would think about is maybe a firm should say, okay, here is, I don't know, an agentic system I want to build for my supply chain. >> Yeah. >> Right? And optimizing the supply chain planning, let's call it. After all, supply chains for an industrial company or for anyone is, you

[07:46] know, or a data center company like us is pretty critical stuff, right? Which is it's it's also idiosyncratic based on what is unique about us and how we operate and what have you. So, you I would say that one of the things that is least talked about, although it's very critical for any agentic or AI system, is the outcome or the eval that matters for any such business process, right? Uh and so, I would start there, which is what's the outcome? Uh and my ability to grade that outcome. >> [snorts] >> Um and start with that and then say, "Okay, I welcome all models." >> Yeah. >> Then, the important thing is the the the all part, which is I don't want to be locked into anyone model. >> Totally, yeah.

[08:31] >> Right? I'm happy to have all the frontier models. I want to use them to see where they are on the eval. >> Yeah. >> Then, I want to be able to use my own context, my own data, in fact, my own traces to maybe even take a much more open weight cost-efficient model or a fine-tuned model which is not using general intelligence, but is using the intelligence I need in order to basically hit the same eval that a frontier model had. >> Yeah. >> That is the empirical way I think companies are going to go about getting value because what that achieves is one, you're grounded on what the result is. Second is you are also ensuring that you're not locked in to any one model. >> Yeah. >> And that means you're in control of your

[09:17] token economics long term. That is how I think companies can decide, which is, you know, really be outcome-oriented >> Yeah. >> uh and continuously optimizing. There's no such thing as, "Hey, this model and this model family is the only thing that I am locked into." If anything, you're locked into one thing, which is the monotonic improvement of your outputs. >> Yeah, I think that's that's really really well said and and I actually hadn't occurred to me before is by simply having the benchmark, you get all of this control cuz you can see how things relate to one another. Um I think you know, in the article you were talking about this idea of like a learning loop outside of your model. And I think this kind of touches on that is

[10:06] like once you have these private emails, which you know, they're very unique to the company, you can start to make your systems better. I'm curious out of the companies that Microsoft works with, how what are the sort of initial foundations of learning loops that you're seeing people start to build around their AI systems? >> Yeah, in fact, one of the biggest things that we are trying to get done is like after all we built uh you know, Office 365, Microsoft 365 as this multi-tenant SaaS service in which some of the most important data {slash} context of a given enterprise is present, right? Which is if you think about it, all their communications uh data and information, their projects, their timelines, their calendars, their

[10:56] documents, all of that >> Yeah. >> is the unstructured database uh where all that tacit knowledge. In fact, it's sort of not only that the the unstructured data is there, but it's the pointers to all the other of the business process, right? Because humans are communicating, creating documents, all this unstructured data is in the context of even structured processes, right? And so, to me, taking that and really in fact, we describe it as Work IQ, we want that to be the context layer along with any model, right? So, when we even built Copilot, that's kind of how we built it, right? Which is Copilot takes the Work IQ context, mixes up with any model, and then helps you with the outcomes you want and the intents you

[11:43] want. >> Yeah. >> And we are even exposing that, like so for example, you know, one of the things that we did even at Microsoft was to take the HR onboarding process. >> Yeah. >> The HR onboarding process has a system of record somewhere, but a lot of it also happens in email, in calendars, and and what have you. It's much So, if we can take that context, and then create an agentic system which streamlines the H, you know, onboarding process, and it's token efficient, right? Because the context is being used for both input tokens, um being you know, being input properly, and then you're not using frontier models to generate and regenerate output tokens that don't need to be regenerated. >> Yeah.

[12:26] >> And that to me is a great way to in some sense create a learning loop that continuously gets better before we even have come to fine-tuning a model or what have you. >> Yeah. >> Then on top of it, you can, right? You can You can also have an all >> That's the layer they stack. >> Exactly, right? And you can always like, you know, whether it's you know, I think that that's sort of what I think people will go is have evals, have context, >> Yeah. >> then have models all hill climb, and as the models are hill climbed and you say, "Oh, but I'm I've got all this new trace data. >> Yeah. >> Let me take all that, tune a a more a computer-efficient model >> Yeah. >> for this particular outcome, and then

[13:07] keeps doing the cycle continuously. >> And and I I I think one of the points you made there is is is actually, you know, kind of a centerpiece of the the whole original article is this idea of, you know, human capital and token capital. And when I and that they're very intertwined and interlaced. And and when I was looking at some of the you know reactions to your post, one of them that caught my eye was like this is like an incredibly positive some take on AI, which is actually not what the narrative is usually around around AI. It's that tokens are going to eat into what humans do and you know, it's a little bit fatalistic that like AI is just going to do everything. But your view is that you

[13:49] know, humans are actually essential to making you know, the token economy grow and these models actually do effective work. So Evals is is one way, but you know, when you're thinking about these these learning loops that are being built, how are how are how is Microsoft thinking about designing learning loops that are very human-centric so that they learn >> It's a great thing, right? Because at some level, you know, I I go back a little bit to Uh first of all, I think we should be clear-eyed about the change to the knowledge economy and the knowledge work. >> Yeah. >> And so therefore, even the displacement with something as disruptive as this. >> Yeah. >> But if I look back, um let's say in the

[14:33] early '80s, long before you were born. >> [laughter] >> Yeah. >> Um if somebody had come to me and said, "Hey, 3 billion people are going to wake up and they're all going to be typing every day." And we would have said, "What are we talking about? Like why would the world need 3 billion typists?" >> Yeah. >> But But we did invent a complete thing called knowledge work. And it is turns out that all of us in different forms, right? Whether you're an accountant or a supply chain analyst or a coder, uh we're all doing knowledge work, right? So we kind of figured out how to use these computers uh to accelerate what was work that was being done in different ways. >> Yeah. >> I think we're at the We now need that

[15:17] same new reinvention of knowledge work. And I I mean, take even coding. Right? I mean Let's you know coding for the longest time even for you and for me it was code as input. >> Yeah. >> Suddenly code has become output. Right? So like you know in some sense we know but but at the same time I do think that you know what does a coder do today? I you know one of the things that one of my colleagues you know thought me and he's in fact written a skill is called cognitive coverage right? Which is the human is still on the loop. Just like test coverage in the past. >> Yeah. >> Humans are needed to have cognitive coverage over machine output. >> Mhm. >> So that you know what the machine produced. You're able to interrogate it. That's a

[16:05] job. >> Totally. >> That we will in fact value. That you will then need skills to do. >> Yeah. >> That you will go to school for. >> Yeah. >> And that will have wage support. >> Yeah. >> It may be different than what a software developer's wages in the past were. It may be more or less the same who knows. >> Yeah. >> But there is a job to be done. >> Totally. >> And so one of my things is hey let's now get into this thing about where are all the human capital interventions to create more token capital. >> Yes. >> And I look around the empirical evidence is there will be no chance of creating any token capital without human capital. >> Yeah. >> And so we are talking about as if this just came from nowhere.

[16:49] >> Yeah. >> Even if you go to you know let's let's say we're all AGI paired we're going to hit RSI. Even with RSI the question is you know unless we're talking about us as humans losing control. I look at it and say the only thing valuable in the age of RSI is the human capital that's controlling the AI. So I feel like that's the imagination we should unlock and starting atomically at the firm level and saying how do I really design my organization now where the human capital, its ambition, its agency is in fact driving the accumulation of token capital. >> Yeah. >> And if we framed it that way, it's positive sum for the individual, for the firm, for the economy, and by the way, we'll be fine with tokens doing a lot of

[17:41] the work. >> Yeah. And you've said this before, which is a lot of it is about finding the right hill to climb. And that'll always be a very human thing. And when you feedback to these AI systems and you define what good and bad looks like, out pops that private email that you're actually trying to go and optimize your token spend towards. But one thing I was actually curious about was, you know, there's a big supply and demand mismatch right now between how people are using these AI systems. And one thing I I was just kind of thinking about this last night, so I wanted to ask you is, is in a world where, you know, demand is far outpacing supply, but there are huge companies like these model labs buying up a bunch

[18:27] of capacity for compute, and compute equals intelligence, do you worry about a world where there may be any like unequal distribution of intelligence across these firms? And what are the ways that people can start investing in their future today? Is it like buying capacity to actually go and run these models? Is it working with a hyperscaler like you guys? >> Yeah, it's a great question. I mean I mean we do have definitely because of where the demand has come from and and then the supply does require time to get built. So we're definitely in the throws of a what is a massively supply constrained market. But I would say like if I were Uh, whether it's an AI startup or a large, sort of bank. I would start with, "Hey, how do I

[19:17] create my own hill climbing loop system?" And that learning loop in which every day I'm sort of really creating that private eval benchmark, right? That is our and my IP. What's that context layer that is optimized in fact to use tokens most efficiently? How do I then use even uh, the models that exist today uh, to help me uh, climb up that benchmark? Uh, and then maybe take even there and say what's a way for me to fine-tune a model, right? So, that's the place to start. And and that way then you're participating. And I think that if we did that I think that some of the supply challenges will over time get solved. I think that this concentration of compute in, you know, is sort of by definition

[20:10] of physical constraint that over time will have to solve except for the network effects. So, that's one of the reasons why I'm advocating for this ecosystem piece is, "Hey, the everyone who's a participant in the economy has to recognize the network effects of intelligence are different than even the classic network effects of digital technologies." This is not just about some tech people competing with each other. It's much more invasive. Yeah. >> Intelligence is a different ballgame, right? >> It's a very different ballgame. And so, as long as you understands it, then I think every firm will figure out how to participate in it versus just be a bystander renting a model only to be locked in and never have the even the

[20:58] freedom even when compute is abundant to do anything, right? Like that's why I say like one of the things I wrote in their article was you can always buy a tool. You can even outsource a task or even a job, but you can't outsource your learning. >> Yes. >> Right? If you do outsource your learning, then why exist? >> Yeah, and and I I I like the comparisons you were making between like technologies in the past and uh you know, what this new AI paradigm is. Uh you know, you you you stepped into the role of of CEO, but you've always kind of operated with this like founder energy. People call you a refounder, right? I'm curious like when it comes and and and the hallmark of an amazing leader is sort of leading companies

[21:41] through these paradigm shifts, much like you did with the move to cloud. Uh when you're thinking about this new paradigm shift to AI, how are you thinking about practically, maybe even within Microsoft itself, changing the way the company operates in order to really catch the wave and make sure that um you guys are building the right things and and sort of >> Yeah, I think see the magic of founders are they create something from nothing, right? I mean, that's you know, what by definition it's sort of full risk on and if you're successful, that means you've created something that didn't exist at all and was conceptualized and then executed on and that takes that energy and and that completeness of thought and the

[22:22] relentless pursuit of it. Uh and for companies that have been successful, uh right? You sometimes forget what made you successful initially and that's why I think uh this refounding as a concept is empowering, right? As I say to for mere mortal CEOs like me who come in later after having, you know, you know, being in a company that was super successful, but not being a founder. >> Yeah. >> But the point is the opportunity for anybody, whether you're a CEO or part of the management team or even an IC in this company, is to be part of the learning culture of this place and be part of that refounding team, right? That's I think a great empowering thing. >> Sure. >> Um and the reason why you have to do

[23:06] that is, you know, you remember like our industry in particular always has been pretty harsh around technology transitions, right? Either you make it or you don't, right? You So, you can't say, "Hey, I'm building um mainframes in the age of mini or minis in the age of the PC." Uh it just moves. And so, when it moves, if you're not on >> It won't wait. >> Yeah, it's sort of sort of one-way doors, right? And so, you have to sort of move with the technology secular change. The At some level, I would say that is the easy part. >> Yeah. >> This next hard part is the business model shift. >> Totally. >> Right? Because you fall in love with how you made your margin, you succeeded in the mainframe or in the mini or in the

[23:51] PC, and suddenly something new shows up and you say, "Wow, >> Yeah. >> this thing doesn't come with the same margin profile, and I'm so yet I'm supposed to love it." And so, that's kind of it's, you know, and the and the street and others will not give you permission, right? Everybody in this you know, in the the street will talk the long game, but then everyone is short. >> Yeah. >> Uh but that's kind of what you have to earn their permission, figure out a path. It's always hard. >> Yeah. >> Uh but so, that you have to conquer the business model. And sometimes the change will even come like this AI change, where even the way you produce the magic, >> Yeah. >> which is human capital using digital

[24:28] tools to create more software. Like that's you know, for the first 50 years of Microsoft, that's kind of what it is, right? We had to deal with a lot of changes in technology, right? From the Altair to the cloud, there was a lot of change. >> Yeah. >> But at some level, the production function was just create more software. >> Yeah. >> But suddenly now software has gone from just being pure input to just output. >> Yeah. >> And so, that production function change is also something now we not only have to deal with what is going to be changes in technology. Uh, the good news here is we were early to it. So, we are I think we are good at you know at it. >> Yeah. >> The business model is actually pretty

[25:09] conducive, right? After all the cloud is a consumptive business model and AI has got marginal cost. So, it's all a consumptive business model. >> Yeah. >> So, the bigger change for us this time around is going to be that last change, which is hey how do we ourselves at Microsoft and you know completely change the way we work across every function, not just software, right? Because you can't change one function because you run into the Amdahl's law, right? Which is you got to have the entire organization not have impedance mismatch and work like an AI native company because an AI native company is born today, they will think about their human capital, token capital in the most effective compounding loop. And we have

[25:52] to now go from our old to the new new fast. >> That makes total sense. And and I think you know Microsoft enterprises run on Microsoft. You guys have built this whole productivity stack on top of it. And I I I know you've said in the past that there's a lot of that that is still very applicable in the in the age of AI and people have asked you like you know, what is the durability of software and whatnot. Um, I'm curious if if if you could talk to like how does software in today's world and people who build software, durable software you know, we've seen the markets say oh software is dead, but really like people are still using these things. Um, what do you think happens to software companies and the products that they

[26:40] produce? Will they flourish in the age of AI or eventually people will just build everything in house? >> No, I think that the good way to think about it is perhaps if you're starting something new. >> Yeah. >> It's fascinating, right? Which is if you believe the firm is going to have human capital >> Yeah. >> and token capital. >> Mhm. >> You then say, "What's the interface between the human capital and the token capital?" >> Yeah. >> And then say, "Wow, if the next big hit thing that we produce which is a fully embodied agent >> Yeah. >> that needs to interface, has to show up in Teams or some other communications medium." That that means there is some need for some software. >> Totally. >> Right? So, that's kind of one. Like oh,

[27:26] even Excel is fascinating, right? Which is Let's say in the model created us most sophisticated supply chain or financial model. >> Yeah. >> But I as a human now need to sort of make sense of it. Uh then the best way for that AI to generate the artifact is to produce a great Excel model, which by the way is just not a grid, right? It's got all sorts of positional >> behind it. >> got all the positional information. I as a human know how to parse the rows and columns. I can do summations and so on. And give me an AI agent even right next to it as a side car that can drive the model. >> Yeah. >> Right? So, I feel like software still has a lot of value. And as long as the but the pricing adjusts, right? Which is

[28:17] at some level if you are a very high-priced piece of software which was delivering low value >> Yeah. >> people are going to ask the question, "Should I just generate it?" >> Yeah. >> Uh generating it is one part. After that you got to keep up with it. You have to make sure that all the cyber stuff is being taken care of not one time but continuously. But that said, I think the physics have changed and so therefore no one can just sit on what they produced. >> Yeah. >> Uh and customers will question what's the value of all of this. Uh but at the end of the day, as long as there is humans and tokens compounding together, >> Yeah. >> the need for software is going to actually explode. I mean, another

[28:56] example, like we're seeing it. We're seeing it in GitHub. >> Yeah. >> Uh painfully so. Uh and also in M365 where agents as they explode, what was systems that were built for human workflow, >> Exactly. >> now have to completely like I mean like I mean think >> Yeah. >> the Git protocol >> Yeah. >> itself was built for a human workflow. And so now we have to think about wow, what does, you know, a work tree mean? What does agentic workflow mean? Uh with code. Same thing with Work IQ. Work IQ cannot be on the same path as my dial tone when I'm in Outlook. >> Yeah. >> So we literally are creating a complete different middle tier such that the reasoning loops of AI that are also interrogating the same data are not in

[29:45] fact sort of crowding out what is the human interaction uh with the data and what have you. >> I love it. And And I just maybe the last question cuz I know I know you have to run is as Microsoft thinks about, you know, the next the next couple of years of this AI shift, there's sort of this question of what is first-party product? What is platform? And then what, you know, as the dynamics and physics are changing, we should expect companies to just go and build entirely themselves. Um do you have some rough sort of guidelines for the types of products that Microsoft's going to build, the types of infrastructure they're going to support, and what, hey, you know, companies, you should just go and and

[30:26] tackle this head on and it's good for you in the the long run to do that? >> Yeah, I mean, I think, you know, the way we've always thought about it and I at least the the way I've conceptualized these ecosystems are is like, hey, you build various layers of the stack. But each layer of the stack, one our case, should compose, but it should be extensible, right? So, what I'm like even take something like Co-work, right? Co-work for us is one of the modalities. We have chat, we have Co-work, we have autopilot or scout, and so on, which are all the modalities of Co-pilot. >> Yeah. >> But the point is all of those also got extensibility, right? So, you can build an agent that is discoverable inside a

[31:08] chat or inside Co-work can use it as a tool and so on. So, that's kind of So, in other words, we want to build every layer of the stack as something that is ecosystem extensible. >> Yes. >> model diverse >> Yeah. >> Right? Those are some of the principles we will have, which is at the end of the day we want to make sure we are making each layer competitive and at the same time making sure that the customers or third-party developers have choice. >> Yeah. >> And not because, you know, competitively that is always the best answer for us. But in the long run, what we've learned is by interoperating by having this positive sum construct >> Yeah. >> we've been able to grow the market and grow our share.

[32:03] >> Yeah. People want choice and they want People want choice and >> sort of in the long run it's much better and more stable going all the way back to the beginning >> Totally. >> which is if I said, "Hey, look, let's not Let's have a frontier >> Yeah. >> where there is a real ecosystem participation. Because if you do that, that positive sum is got lot more stability for the economy. >> Well, thank you so much for joining me. It's I really appreciate your time and yeah, looking forward to working more together. >> Likewise. Thank you so much, Yash. Thanks. >> [music]