Blitzscaling with Microsoft CEO Satya Nadella & Greylock's Reid Hoffman | Greymatter

Welcome to Gray Matter.

This episode features a fireside conversation with.

Microsoft CEO Satya Nadella and Greylock partner Reid Hoffman from Greylock's event Blitzscaling. The two luminaries share personal stories and learning lessons from their experience. Blitzscaling companies, including Microsoft, LinkedIn and PayPal. Enjoy. Thank you very much for coming. One of the things when we said, well, who should I kind of come and talk to Silicon Valley about, both for what's going on in interesting tech and for Blitzscaling, my very first choice was Satya. We worked this out and it was awesome. The real thing is to get to interesting content and we're gonna actually ask each other questions. Like part of the thing when we were talking about it is like Satya was like, no, no, I, I'd like to be interviewing you. And I was like, no, I'd like to be interviewing you. And so we said, okay, we settled and we'll both ask questions. Because I have the pleasure of the first of the mic. What I will actually start with is, I think a question close and dear to Satya's heart, which is that Blitzscaling applies in a number of circumstances, including changing the culture of a company as you take over. And we'll get to like a little bit later when I'll ask some questions about Microsoft being one of the original Blitzscaling companies and what did that look like? But let's start with a cultural question, because I think one of the things that I've had the delight to see in collaboration with you and the board is literally the rapidity at which the culture and the way that the growth psychology, the approach to market, the approach to collaboration, the approach to this is how we operate. Changing at speed. Right. And in some senses would be a surprise because you're a long term Microsofter. Right? You would think you would like, well, I've been bred in this other successful culture, but I have to change it. So say a little bit about how you thought about that change and then how you thought about scaling that change quickly and powerfully through the. Oh, wait, I have something else to do first. I was going to go, yes, you're right. Sorry, I was starting with my very first question. Different question. Paul Allen. So obviously it's a huge loss. He has not only been a powerful industrialist but also a massive contributor to Microsoft. Why don't you say a little bit about kind of what he meant to you and also a little bit about what he meant to Microsoft.

Absolutely.

It's a very sad day for us at Microsoft, Paul is someone I've gotten to know a lot more, in fact, since becoming CEO. He's one of these, I'd say, understated human beings who did a lot for us, obviously, as a company, having founded it for our industry and the community in Seattle, he's done so much that he'll be missed sorely. And more than anything else, I think Paul is one of these people who was always curious. You know, you meet him, you would never think of him as, wow, this is the guy who actually founded Microsoft. He was always asking me on what's next? And I always found that to be an amazing attribute of his and his curiosity and his generosity for everything that he did will be surely missed. In fact, it ties a bit to your question because you asked about our culture and I was talking to a group of founders, and what I said to them about this culture thing was I consider myself as a mere mortal CEO who took over from someone who was not a founder but had founder status. Steve, for all intensive purposes, Paul and Bill started the company. And in some sense, Steve and Bill scaled it. And I made a conscious decision. Even I grew up in the company, and I made this conscious decision that I'm going to make culture something explicit that I myself will talk about. It's interesting. I found that at least in companies led by founders, culture is sort of implicit. A lot of it is bound up with who the founder is because they've obviously scaled it from nothing to what it is. But especially for a CEO taking over, I recognize the need for me to have much more of a conversation. Being an insider, I could call it for what it was, right? In the sense the good and the bad. I represented it. So there was no denying it. It's not like I could say, hey, I came from the outside. In an interesting way. I had credibility to criticize Microsoft without it appearing as criticism. That's at least how I think the advantage of an insider, because there's always the debate about the outsider versus the insider. The insider does have an advantage because you can say it in a way and do things in a way that it's not viewed as criticism. If anything, it's viewed as being intellectually honest. And I always used to say, at Microsoft, we. We had this amazing advantage called Bill Gates, and then we had an amazing disadvantage called a bunch of people who were roaming around Microsoft thinking they were Bill Gates. And that was really the gist of it in the late 90s, I distinctly remember that, the hubris that it set in, that distinction between what you Contributed versus the rocket ship you rode becomes sort of and a blurred for a lot of people when success comes, and especially when success comes easy. So that's one of the reasons why I went on this search. And weirdly enough, the place where I found the metaphor for at least having that cultural conversation was a book I'd read. My wife had introduced me to it called Mindset, like maybe four or five years before I became CEO. And Carol Dweck at Stanford, she has the simple metaphor which is if you have two kids in school, one of them has got a lot of innate capability, but is a know it all, the other one has less innate capability. Is a learn it all. You know how the story ends. The learn it all does better than the know it all. And it's true for students in school, it's true for CEOs, and it's true for founders. And it's true. And so we took that. And sometimes people will say, hey, this growth mindset is interesting. Satya I found the 10 people at Microsoft who don't have a growth mindset and that's not the point. The point is not to go look around for 10 people at Microsoft and who don't have the culture. The thing that I love about this growth mindset is the ability to confront fixed mindset. In other words, I have to be comfortable admitting that I have a fixed mindset. And every day what you do is your best job at trying to confront it and if anything to your scale point. Right. If you go back to the blitzscaling thing, I don't think this cultural renewal at Microsoft would have happened, but for two fundamental reasons. One is we did not think of it as a one time transformation from attribute ABC to attribute xyz. It is going to be a continuous journey and in fact I'm very, very careful not to celebrate some destination reached. Second is it was not viewed as new dogma from a new CEO because if this didn't have that attribute of foundationally appealing to the human in us, all of us, then I don't think it would have worked. And so the fact that you don't do growth mindset or confront your fixed mindset to be a better employee at Microsoft, you do it because you'll be a better parent, a better partner, a better colleague and a better leader. And the fact that it speaks to your entirety of your life is what probably really helped us.

And what were the first couple of moves that you did to start trying to get it to spread what was a set of things of here's how we need to shift course and go on this continuous learning path.

The biggest thing in retrospect, I understand the power of this a lot more. We are 130,000 people in the company and I've learned a lot about what does it mean to be CEO? Because nobody teaches you, hey, here is how you be a CEO of 130,000 person company. You suddenly get plopped in and here you are. And at that scale, the simplicity of what your overarching framework. I mean if you look at it, there's not a single meeting at Microsoft that I don't somehow manage to go to our mission and our culture. And I've done it now close to five years and I am so disciplined about it because that is my unit of scale. The day I stop doing it, the entire motion will fall apart. And as you scale, stop inventing new things that you talk about, especially around your sense of purpose and mission and culture. And then we had practical things. We picked I would say three key areas because it's an abstract thing. What the heck does it mean to confront your fixed mindset? We said, hey look, it's easy to do. Take customers, listen to them, don't go, just pitch to them. But get behind. Whether you're an engineer looking at a log file or a salesperson, both sides can in fact go deeper on what are the customers really trying to say, what is behind those words? Take inclusion and diversity, talk about that area that every day you can practice, you go into a meeting, you have a diverse group of people and people who come from different backgrounds are not able to get their word edge wise. Happens quite a bit at Microsoft. And so then you say okay, how can I as a leader or as I as a participant change the culture as we speak? It's the living breathing culture, not the abstract one, the other one. We've had our challenges were we had gotten organized into these beautiful business units and we would metric them to hell and we would measure everything and it was great. Except once they're all rear view views, right, which is they were all categories that had gotten created long time ago and they were not as relevant. We needed to mix up. We made a big structural change too by the way. That was one of the biggest changes I did, which is we had a hundred thousand person company other than LinkedIn that we operate and somewhat our gaming that we operate in a much more loosely federated way. It's a functional org and it's not because I love functional orgs, but I needed to do that and in fact, Steve got us started on that in order to make sure that the company was not stuck in an. Org configuration that was backward looking. And we could actually, from silicon all the way to a gaming experience, rethink everything from the product we build to the business model. And so those are all the things that we had to explicitly go do.

Yep. Well, one of the things about keep saying the same thing is I look at that as establishing the frequency. It's almost like, let's harmonize all the same frequency. And I just came from a micro version of that, which is when you're on media tour, you're saying the same thing again and again and again. And you're like, really? Aren't people bored yet? I'm bored, but you try not to show that. And actually the funniest thing I think that made me stronger at that was the Masters of Scale podcast. Because as you're saying, as you're kind of reading a script and you're trying to convey excitement about this, that even when you're reading a script, how do you have people share that journey with you even though you're like reading the script for the third, fourth time as a way of doing it?

But talking about this idea of scaling, though, what is your perspective for a large company like ours? Because in some sense, you think about in your book, you do have the hierarchy, right? It starts with the family and goes all the way to the nation or the globe. But if you sort of had to reflect a company like Microsoft or any one of the other large companies, what does it mean, though? What does a blitz scaling mean for a large company?

I'll open with the definition of blitzscaling, which is the prioritizing split speed over efficiency in an environment of uncertainty. And the uncertainty part is a really important part of this because actually, in fact, if you have certainty, like you know what your customer acquisition cost is, you know, your exact competitive landscape, your unit economics, your ltv, then you can still scale fast, but you know how to do it in a relatively efficient way. You don't actually have a lot of footfalls. It's a bit of a spectrum, but you still have that. So the uncertainty is actually, I don't know what my customer acquisition cost is. I don't really necessarily know what my total addressable market is. I have an idea that I can get to good unit economics, but I don't per se know that. And then speed over efficiency is. Well, actually, in fact, I still need to get to scale very fast. Usually driven by competition, usually kind of the uber circumstance, we open the book with Window and Airbnb, and so usually driven by competition, but sometimes it's potential competition or sometimes it's the economics only really work when you get to a certain amount of scale. So you still need to go to scale and get that early. So, for example, how did PayPal get to a good fraud engine? It was like, well, we had to get to scale of data within the system. And so blitzscaling was important there anyway, even though you didn't actually really understand the economics, because the entire economic curve would change as you understood what your actual fraud cost was and how you can manage that, because the unit economics per charge and then the speed, obviously competition, capital allocation, other kinds of things as a way of doing that. Now, when you get to a large company, there's a couple different ways that a large company could think about blitzscaling. Now, one of them, the reason I open with culture, is because sometimes you say, no, no, we're going to change the way we're going to operate and change this. That's also a blitzscaling phenomena. You have uncertainty. How well will the culture take? Will we be able to do it? You need to do it fast because you need to get everyone rowing in the same direction. Like if some people are rowing this way and other people are kind of rowing this way, you have disharmony. It doesn't work. So speed also matters. And to some degree, your question of efficiency as well, the per dollar or per human, it kind of doesn't matter. We just need to get there. And so that was the reason the culture was one of the fascinating things. I think you're on path to no destination, on path to another one, which I'll ask you about in a moment is Azure, which is this question where you say, well, actually, in fact, this market we have to participate in, we are not there yet. We have to get there, we have to get to a certain scale and we have to be competitively in the fight. Right. Given how far out AWS started, when you're really pushing forward to getting there, that's another, I think, example of a large established company going, we need to play now. In order for a company to do that, I think a couple things have to be true. One is, I think you have to be able to get sufficient backing from investors. Right. I think that you need to have a kind of a we buying into the future of what it is now. It can be a challenging way of doing that. For example, one of the examples we used is we said well, so say you had a $20 billion public company and you had 17,000 employees. Wouldn't you think blitzscaling has passed you? Well, this was Amazon in 2006 before AWS. Right. And so it is actually there. And part of the reason we use this example because in media, when Bezos started doing that, the response from the street was there was literally a magazine cover that said, Bezos should just mind the store. Right. So the response was not like, oh, yes, that's great, we're all going for it. You need enough buy in. Not enthusiasm necessarily, but okay, this seems to be good leader, good shot. We may disagree with you, but we'll give you enough that you'll start proving it. And I think that's one of the other things that you've also brought to Microsoft in this, which is, oh, we believe now that Microsoft can invest in the future and we should be enabling Microsoft to invest in the future because we believe that future will play out well. So one is the street, another one is you can't unless you're in desperate straits. You know, Apple, when Jobs went back, this kind of thing, usually you can't blitzscale the entire business. Right. Cause of the disruption. So usually it's a component. For example, you know, AWS was a group, it wasn't. We're just doing every. We're just, you know, resetting to zero. And then that means that that team needs the right backing from the CEO or you know, the kind of buy in as the derivative down to the investor to say, we understand that a dollar here will have an uncertain dollar output. Because part of what happens in these traditional industries is they go, well, this group goes, look, you give me a dollar, I'm going to give you $1.80 back. Because we know it, we've got it, we're tuning it for efficiency, et cetera, et cetera. And then when you're giving the dollar to the Blitzscaling group, you're like, well, when are your unit economics going to work out? Well, soon we have a spreadsheet and you have to have the credit for. No, no, no, we're going to build into that. And so I think that's some of the traditional company. And so talk to me a little bit about how you made the Azure decisions and how that kind of worked through because that's kind of another example of the key kind of blitzscaling kind of groups within an established company.

Yeah, I mean, a lot of what you said is obviously you didn't have that vocabulary, but I Always sort of say that look, long before it's conventional wisdom, you would have had to take that risk. And you do it because you need to have that courage to in some sense have the speed over efficiency with an uncertain future. And it comes because of the intuition you have of the arc of technology and the microeconomics that goes with it, right? That's where the founders, everyone here has that intuition and that's what gets you to even do the startup. And to us, that's really what was the case, which is we understood, hey, where's distributed computing going? In all honesty, we started started late. And so therefore it was more about being able to, in that case, be a fast follower and then be all.

In your own area too, which is super interesting. The hybrid is something you should go into.

In fact, I'll come to that because now, interestingly enough, here we are now leading. It's kind of an interesting way. But the way in the beginning when we got started with Azure is I remember distinctly in 2011 joining the team, that was our server team and we had an amazing business with amazing gross margins. And they would look at this cloud thing and say, God, this is a nightmare. But yet it was clear as day that this nightmare could in fact finish us all off. And so we needed to just be all in. And it was unclear. In fact, if you think about the street can punish you, right? Which is once they have a set of metrics that they're tracking you on, you better be just on that treadmill. But the thing that I learned, Reid, was even with the street, street, the investors, and of course investors, I think Bezos says this, you get the investors you deserve, but even the investors, you can go and ask for permission, but the only thing that they need is real conviction and real progress quarter over quarter. So one of the things I did was to ensure that we had leading indicators of success. So I had to sort of basically get a set of metrics that were not net profit for sure, that were not even gross margin, but were indicators of something that were going to be long term gross margin or long term net profit, that the investors said, okay, as long as you walk the walk every quarter on this set of metrics, I'll give you the leeway. So to some degree, we went for scale over efficiency, especially measured by any of our traditional metrics of efficiency. Got permission from all the other thing I realized was, again, this goes back to being mere mortal CEO, not a founder, right? I can't go in and say, hey, trust me, is to walk the walk. And I realized that if you can do that, even public markets can be a fantastic source of investment. And Azure now, interestingly enough, maybe three years ago at our developer conferences when I first started talking about the edge because I always distributed computing is. It's not going to be like the back to the mainframe. I mean we started distributed computing will be distributed forever. The question was what is the edge going to look like? And if you think about what we are doing with Azure Sphere, which is basically a silicon design on operating system and a cloud service for a microcontroller to have a compute node to Azure Stack to Azure is one distributed computing plane for doing sort of the new workloads. This is not a hybrid for the old world, this is really the hybrid for the new AI first workloads because compute gravitates to where data is being generated and that's what's driving a lot of our growth. I mean I look at even a traditional company like Chevron or Shell, it's amazing. The amount of compute power we're putting at the edge in their rigs, which is just part of the Azure fabric is a great view going forward.

Exactly. One of the other things that we didn't have a chance to talk about is part of the reason we had Bill write the forward to Blitzscaling was we thought of Microsoft as like one of the actual early original Blitzscaling companies where it got this super compelling and powerful business model and then went to global scale. What was it like being part of that company and then saying okay, now we have to reset? Because I actually think most Blitzscaling companies actually get to a point at some point where they need to reset because you need to start focusing on we're going to focus on efficiency. Too many of these kind of multi thread experiments. And so there has to be kind of now we're going to have much more efficiency. What was that like both being in IT as you kind of grew through the ranks and then thinking about how is it that we reset our ability to earn our market rather than take it for granted.

Yeah, I joined the company in 92 and the mission of the company at that time was put a PC in every home and every desk. Talk about a concrete mission, which even had the math formula right in there. So by even the end of the decade, I would say, at least in the developed world, we had scaled to achieve that and it was amazing. Like for example, when I joined the company there was no enterprise business and the joke was that we'll never be an enterprise company. And here we are, whatever, 25 years later, we are now the quote unquote the enterprise company. But it takes that long to build it out. But we had the inherent belief that whether it was our technology choices that we were making, whether it was an operating system with the intel architecture was going to go into the data center was fit or complete, I mean in 92 you could just look at it and say it's going to happen. We all overestimated how quickly it'll happen, but it happened eventually, maybe five, 10 years after we originally thought it was. But it was locked, the scale was locked in. The thing that I would say though I've learned is the biggest challenge comes at some point. Anything that has hyperscale will eventually taper off. And when it tapers off, your ability to invent the next thing, if it was just a technical shift, you have much higher percentage chance of hitting that just because you will have a big R and D budget you would have spent, you would have covered your base. But if it's a business model shift which is fundamentally destroying the margin that you celebrated all these days, then it's a hard one. So for us, for example, the mobile thing was just difficult because it really changed in two ways. It became hardware gross margin for Apple and it became services on the back end for Google. And here we were with an OEM based business model. And that transition was the hardest for us, which is we needed to reinvent our business model. And that's where I would say we faltered. And whereas in the cloud we didn't, we decided that we were not going to be beholden to the old business model. We were not only going to change how we built products, but we were also going to change the business model on the products.

Makes total sense.

The one thing that when I was reading your book I was thinking look at the number of challenges that we face as of society. In fact, interestingly enough, when you think about that speed versus efficiency. I run into this nowadays around communities. I've gone back and studied a lot about the Xbox Live community and one of the reasons why that community has integrity around a lot of things is because I think we made some very good conscious decisions in retrospect. Thank God we did that to grow it with moderation really turned on in a big way early on. So I think about there are a lot of unintended consequences of scale, especially at a societal level. And it's not just tech, it can be a program by the federal government. How do you think we should Think about scale versus efficiency when there are unintended consequences, sometimes of scale.

So one of the things that we added in, and actually there's an essay which I'll come at the bottom, which will come back to the question, which is, I actually been learning a bunch from you and Brad Smith about how tech companies can think about their position of responsibility within society. And what I realized after we wrote this chapter on responsible Blitzscaling was that there's another article that I should do. So I'll come and talk to you guys about that. We'll may do a little bit of it here too. But the basic thought was the challenge that these blitzscaling companies have, and I'll get to the society part of it is when you're small, it's fast, and it's as fast as you can go. Like, the thing that I learned from my first week in the operational job at PayPal was every decision was made with the time frame of you do not let up from a full depress of the acceleration pedal. Right? The amount of time you had to make the decision was, well, you're going full force and make the decision by the time you get there, which may mean right now. And so you have this challenge because you have to move fast. Competition, getting the scale. And you say, well, where does does responsibility factor in? Where does this thinking about what your social impact is, what your risk factors, especially as you get to scale. And one of the things that is a kind of a goal in writing the Blitzscaling book was also not just to reflect what the Silicon Valley playbook is, but also try to help shape it in some places. And one of the things we realized was if what you did is you said, look, as I know, as I get my organization larger, as I know that I get to more multi, threaded, I can actually start investing in what my responsibilities are because I can start thinking, okay, now we're bigger, we're doing more things, and we're still really running to compete. But we should have somebody, some part of the group thinking about our responsibilities and what our impacts could be. And the three areas that we said every Blitzscaling company should think about. One is, well, do you have serious impact on individual customers? Right. And this is a little bit like the Theranos essay of saying, look, you blood test serious impact, right? No, you can't like fake it until you make it as an entrepreneur. When it's blood tests, right? Not bad, right? Second is, do you have a real impact? Maybe it's not as mortal or as Kind of threatening, but it's a serious impact and it's a large number of people. So if you look at the integration across it, you say, well, actually, in fact, we're affecting a whole bunch of people. And so before it happens, we should figure out what the possible antidotes are. And by the way, nothing can be perfect. But like, for example, if people were coming in and saying, oh look, you've been hacked and you've lost this data and said, look, we've had this whole effort going and we've had it going for years and we've been really trying to invest in it, then it's like, okay, you still fucked up, but you didn't fuck up quite, quite as bad, right? Like it's kind of as an issue. So this one is another area and then the third one is systemic impact. Can you take the system down? Can you make payments, stop working? And in that area you should also say, what are the areas where as we get to large, we're no longer kind of the pirate ship, but we're actually the Navy in which everything runs. We're the infrastructure. What are the responsibilities for being infrastructure? And you should start thinking about it as you get to multi threaded, like as you begin to get to what we call the village stage, which is hundreds of employees, you might say, well, we got somebody, a person or two whose job that is and is doing it. And then more as you go up. And that was part of the thought we had when we were doing this book, is that not only is it to share to other places outside of Silicon Valley what we've learned from here and how to spread that, but also help shape Silicon Valley. And one of the things that I was as I was, because we pushed our publisher a lot, I was literally putting in changes until like four months ago. And the publisher was like, you're not allowed to do that. I'm like, that's indifferent. What's the constraint? The constraint is, well, it has to have the same number of words on every page. Okay, we can still make those changes. And so part of it was trying to get this stuff in. And as I was doing that, I realized that a bunch of the stuff like Brad's digital Geneva Convention and other things were since you guys have been there early and have been interfacing with what the right kind of like, how do you communicate, how do you help set an agenda? How do you go and say, look, actually, in fact, we're mission critical. Infrastructure don't regulate us, trust us is stupid. That's not the dialogue that's gonna work with society proactively saying, well here, here's the kind of things we should think about as a good regulatory framework. What would be the lessons that you'd say from a Microsoft program saying to the up and coming blitzscaling companies, look, here is some of the things you guys should be thinking about in terms of contributing.

I mean, I've been thinking a lot about it. Even in this context of scale, the overarching thing that I would say I've come to realize is day one, while you're building that product or service that obviously needs all the scale, you would be better served to construct, even in its core business model and in the unit of scale, a way to deal with the unintended consequence. So the more you have thought through what are the unintended consequences that will be unleashed when you are at scale day one. And I say that with all the respect to all of the things that can't be forecast at the all right, so the uncertainty is real. But there are things that you could sort of say, wow. And that goes back to something like the Xbox Live. Because the decision those guys made at that time was to say, hey, moderation is going to be important. We want a safe community. And so in some sense they sacrificed a lot of efficiency. It is not like all this AI driven community. They just literally put a lot of humans and said we are going to day one have a lot of more moderation, which is human driven, which is super inefficient. But knowing because the unintended consequence of having a community that was not safe would be problematic. That's the way we are increasingly thinking about our stuff because and somewhat some of it comes with our territory, right? We bought a recent company called Flipgrid which is used in education for giving students a voice. And it's scaling in a very big way. And the point is, we know now we've got to take all the lessons learned. It has to be GDPR compliant. There's are all of these things. The transactional costs of tech are not going down. They're going to only go up. If there's one secular thing that's going to change for all of us is that doing business worldwide is just going to be more expensive. So that means there is an incentive for us. In fact, there's a competitive advantage for everyone in the room here to actually outthink your competition on dealing with unintended consequences. So that's kind of how I think about it in terms of competition no longer Are, did you scale? Do you scale safely? Will be the only thing that may matter. And so that's at least how I'm thinking about it.

Yeah, 100%. And also I think part of what I've seen you guys do, which probably sit down and interview Brad and you and maybe a few other folks, has been this look, there's a way to propose a lightweight set of kind of regulatory frameworks that enhance trust, still create the maneuverability for innovation and aren't trying to just do what people claim classically think of companies and like the digital Geneva Conventions perfectly is it's not trying to disbalance the field in favor of one company.

And that's a very good point.

It's trying to just say let's make it good for all of us.

And that probably is one thing I would say is when I compare Silicon Valley and the maturity of I'll call it the players in Silicon Valley versus other industries, we have still some ways to go because where we think of as our terms of competition, competition versus what is the thing at stake for the industry, we're just so lopsided. We don't get that, hey, if we don't in fact self regulate, we will be regulated. For example. One of the considerations, take all of this AI on the facial recognition issue. The reality is when we look at it, the maturity, you remember the Gartner guys and others would always have these maturity models. That's kind of how enterprise adoption would go. But we don't in scale, we don't apply that. We just take some model that's just half baked and say it's ready for scale. Damn the unintended consequences.

This is blitzscaling, by the way.

Yes, but the problem with that, when you know, if you did it and said oh it was an unknown, that's blitzscaling. Yes, it's completely known.

But when you know it, you should.

That is not blitzscaling. And so that's sort of to me the challenge sometimes we run into, especially for a large company to do that, that's unforgivable. So that's where the Geneva Convention comes in. If you think about it's amazing cyber, we know it economically impacts basically citizens and consumers world over and small businesses because large businesses can spend the money. Tech companies are more competent in dealing with their own network vulnerabilities and what have you. So therefore that is where that Geneva Convention came in. And I think a lot more cooperation, industry led initiatives I think would be important. And regulation is inevitable. You might as well, engage on the policy to make sure that the regulation is smart.

Yes. Has the right shape and is adaptive over time because people say, whoa, everything's going to change. Fine, start very light. Start with like that, by the way.

Is the Chinese model, from what I can observe. There's a lot of talk. I know Kai Fu has written a recent book. It's not as if China is open season in terms of data. They are building their own set of regulations that is more along your lines, which is in some sense the state policy there seems to also understand blitzscaling their policy.

Yes. Well, this is the thing I was just about to also go to, which is most of the time when you get within a social context, you say, okay, not companies, but social society. When should society be thinking about blitzscaling? And the challenge is one of the great goods you want generally in society is predictability and reliability. And so generally speaking, you want to only move into areas of uncertainty when it's really important. It's being driven by some factor and the most obvious visceral one is war. It's like, okay, we just have to do all this because it's desperate. But for example, big transitions which can be the. Okay, we'll actually, in fact, we have all of this kind of industry transformation and we have all these issues around middle class jobs or opioid academics or so forth. And that's when you might start saying, well, actually, in fact, the speed at dealing with this problem because it's a systemic problem, we just need to understand that we're going to take inefficient shots on goal, right? And part of it, and this is one of the things that I actually only really just started having this particular thought as I was answering the questions on the media tour last week, which is part of that, I think also needs learn. One of the things that we didn't cover quite as much in the book is experimentalism, right? So like part of when you can think of inefficiency, you say, well, we take 10 shots, like, we're going to try this one, we're going to try this one, we're going to try this one, we're going to try this one and we're going to look at all of them to say, ok, this part of this is working, this is working. Ok, let's triple down on those and move the other ones out. And government's usually pretty bad at doing that, so we need to figure out how to do that. The Chinese actually, I think do this in some good ways, which is I'd been thinking about what would be the US examples of the Chinese special economic zones. Like what would you do is you say, well, we need to run some experiments on how this stuff's going to work. Well, could we say, look, there are certain areas that are in somewhat economic disrepair. We just need to be juicing them, whether it's Detroit, New Orleans, and say, okay, let's try to get this back, let's make this a special economic zone and let's work in the following ways. And what I'd realized is an essay that I had written and published three or four years ago, which was around special economic zones, saying, look, what you should do is you should allow unlimited immigration here with safety checks and everything else with jobs. If you're investing in this community, if you're getting the economy of this city region going back to going as a way of doing it, you could do it through, for example, a payroll tax. So you have to say, we're paying U.S. rates, it's competitive U.S. rates, and we're actually even paying more in order to do this. And I think those are things that we need to start thinking about. How do we add to the societal corpus, Right? And I think the Chinese, the Singapore model, I think those offer some of the tool set that maybe we could figure out our version and apply them. And maybe the right first one is like special economic zones.

That's a question. Let me ask you one other thing. In your book, I know you sort of explore case studies out of both Silicon Valley and China. What would you characterize as the core difference in the blitz scaling attributes between Silicon Valley and China?

So China has some advantages that it's deploying very well. One is raw scale of employees. So for example, when Tencent said, okay, here comes mobile and we have this QQ that we're making all this money on from the SMS fees and so forth. So classic like business model disruption possibility. They literally started multiple groups in multiple cities and said, all of you have the mandate and you're competing with each other. Right? This is kind of strange from an American corporate perspective, which is, well, I got these three different groups and they're, they're all competing with each other. So it's like a battle royale survivor game. Well, whichever of you make it out right of this mud pit.

By the way, Microsoft, when I joined in 92, we had that, we had OS 2, Windows NT and Windows all three. And I distinctly remember, in fact, the first meeting I went to was Steve Ballmer. It's OS2, OS2, OS2 the next month he said, it's Windows, Windows, Windows. And there was blitzscaling.

Oh, interesting. Okay, that's another conversation we'll have. Yes, OS2, some people probably still remember what that was. And so one part of it is the pure talent, amount of talent, so like number of groups competing, et cetera. Another one which I actually learned from got a real up and close and personal with LinkedIn China. One of the ways we were navigating China, which you still have, which as you know, which is this Chinese only product called QI2 because it was one of the ways that we thought we would also show that we're investing in China, that we're actually experimenting with things that were local to the market in addition to kind of the global LinkedIn network as ways of kind of running a dual process. And so I was sitting down with the team and they showed me a set of wireframes about what they were going to build. And they said, we're going to have this done in eight weeks. And I was like, I've done this investing game a little bit. I was like, no, there's no chance you can do this in eight weeks. I've seen so many companies here in Silicon Valley and Silicon Valley moves faster. I'm like, no, no, no, you don't have to over promise, just tell me when. He's like, no, we'll have this done in eight weeks. He said, look, it'll be really helpful for the team if you come back in a month and you visit us so that we can ask you questions and a morale and so forth. And I said okay, fine, I'll change my calendar around, I'll be back in a month. And I got back in a month and what the team had done is they'd taken all the product team, all the engineering team and all the design team and they had checked them into a hotel. Two weeks on, two days off. Two weeks on, two days off.

Bill would say he used to watch the number plates of the people when they came in and when they left. That was his unit of scale.

Yes, exactly. Well, it was his version of what Lejeune calls 996-09-00am to 9:00pm Six days a week, you're discoverable at your desk. That was when Xiaomi was 30,000 people to give you a sense of how it operates.

So he doesn't believe that you can go into the house and check in?

No, no, it's discoverable at your desk. Right. And by the way, Chow Chow is doing that as well now. But it's 1010. It's 10am to 10pm Six days a week. Right. At whatever 40,000 people or whatever number of people they have. Anyway, so the LinkedIn example, I went back in four weeks and because he didn't tell me he was going to do it this way, because I might have believed him, I was literally like, oh, my God, you're halfway there. Well, yeah, you're gonna be there in eight weeks. And it was because of this intensity of literally when they were at the hotel, it was breaks, 30 minute breaks for meals, you know, 15 minute breaks for like a little walk, a smoke, et cetera, et cetera, and then work. Right. You know, and for sleep. And this was the LinkedIn JV in China. Not only just the kind of raw others. So that's another thing that China folks do. And then the other thing that they're actually doing, which we kind of Silicon Valley.

Just so that. I know what you just said, Reid, you basically said we are lazy.

Yes, yes. Well, compared to the Chinese. Yes. Right.

Fascinating.

And part of that, I think culturally that most of the folks who are around our age a little older remember the Cultural Revolution and remember the starvation that came from it. So that's literally the definition of hunger. Right. They're like, this is our golden ticket. We're going for it. And that's important to realize. And so part of the whole point of writing Blitzscaling is we don't advocate that U.S. or European, other companies can necessarily do the European model. It's possible at the village stage, maybe some, certainly the tribe, certainly the family stage, but it's not, as you get to city and nation, which it is there. And so there's a stack of those things. Now, what they are further behind on the curve of is we've been refining management, we've been refining how do you manage efficiently, how do you do these things, including in Blitzscaling. So it's kind of a compare and contrast. And the lesson usually in blitzscaling is play to your strengths. That's right. Like, yes. Lightly try to correct your weaknesses, but don't try to. Like, if you say, great, this is where you're strong, we're going to play the game on your terms. Bad. Yeah. Now, the last thing I'd say is part of the thing that I don't think anyone here would have this illusion. Oh, the Chinese copy, et cetera, et cetera. Oh, they're inventing all over the place. Right. Like, I get this question from Washington all the time, like, well, they're just copying Our things like no, no, no, no. Like I pay attention. And one of the things that is actually the whole digital goods market, the business model of digital goods is coming out. That's a Chinese like that came out of the WeChat, the weishiing environment. And that's coming back. And that's something that we're starting to learn as an interesting business model. Because which of the business models that scale are actually. It's a fundamental part of the kind of the blitzscaling is saying these ones work, these ones don't work as well. Now, just so that people realize it's not just a tech thing, I'm going to actually go to another blitzscaling thing which is there's a company that some of you may be familiar with called Zara. And Zara is fast fashion. It's a fashion manufacturing. And it's not manufacturing in China for low cost, it's manufacturing in Spain. But what they did is they said, let's close the loop, let's make it really fast. We will use our stores and our salespeople as immediate intel and that intel will go back to the designers and go back to the robotics. There is some robotics in the factories, but not only robotics to produce and ship new clothes to the stores such that it's a two week turnaround. Right. That's a way that you say, look, we now need to work in a world of data, in a world of learning, in a world of faster clock and faster acceleration. It's hessian as a way of doing that. So it's totally doable across the, across the industry. And so now it may be that we do it more within our pure tech companies because you know, we have maybe margins, we have a better understanding attack, we have a better understanding of how do you build products on data, how to use data in competitive Intel. But just as software is transforming the world, it's essentially going, yeah, right, the whole area. And so that's part of the example that we use in the book is it's not just us.

That's a great one. That's a great one. I mean one of the fun things for me is to see today pretty much every company out there is becoming a digital company. One of the foundational challenges is as much culture, it's as much understanding the physics of digital network effects changing in some sense all these other industries and you sort of run into this which is structurally, sometimes companies are organized for a very different factor of production. But now you have a new factor of production and you need to Reorient. And right there lies. It's not as much, in fact, they're willing to change their business model, they're willing to change their product, but it's that structural change that allows them to get to those network effects that's sort of the real either accelerator or impediment.

And since we have a bunch of geeks here, I'm going to ask what is hopefully a quick geeky question just for fun, and then we're going to open it up to questions for from you guys. Quantum computing, something I still haven't quite gotten my brain around. Right, you've been focusing on this more. What do you think the way that people should pay attention to quantum is? And what do you think the potential is there?

So the way we got started on this was there was a CTO at Microsoft called Craig Mundy who's still in fact very much associated with us, who multiple years ago got this program started at Microsoft. The intuition for it or the motivation for it quite frankly, is a lot more easy to get today than when we got even started. Because if you're spending 10 plus billion dollars a year of capital on essentially building compute, you are going to be very sensitive to what comes next in terms of compute power. And there are all these computational problems, whether it's that enzyme in food production or the catalyst that's going to have to absorb all the carbon when we're looking for the miracle around climate change, which are still computational problems that can't be solved by classical computers. So the intuition for why or the motivation for why we need quantum is right there. The approach we took is controversial because what we said is, look, it's not even about achieving that quantum supremacy as a milestone. Because if you just did that when it was not a general purpose quantum computer, you wouldn't have really achieved say what the PC represented or the mobile phone represented as sort of a democratization and the birth of a complete new era of computing. And so we're going for it, which is we're going for building the general purpose quantum computer. And the challenge of course is how do you stabilize these qubits? So there's this guy who is a Fields Medal winner, works at Microsoft. He came up with this concept of a topological property that's a way to stabilize these qubits. So we've bet on that. Now the challenge of course, is to go find the physical instantiation of topological matter. And somewhere in the 1940s or 1930s, I think a paper was written by a theoretical physicist that There are these particles called particles that will exhibit these topological properties. Lo and behold, we have these world class physicists who someday will win a Nobel prize because we are close to stabilizing those particles. So essentially it's math instantiated in some physical properties that we discovered. And then of course you got to build an entire computing stack and there's all sorts of sophisticated things. What's the controller to the quantum computer? So we are in it. We're spending a lot of money. You and the board approve it?

Yes. And by the way, it's still mind bending.

The nice thing is this goes back a little bit to the blitzscaling thing is one of the other, at least terminology we use is what we call no regrets move right. Even let's say a lot of these things take a lot longer than what we have envisioned. Are there byproducts of what we are doing that can in fact be monetized, used in other contexts? A great example of it is we now have the quantum simulator on Azure. And the quantum simulator on Azure, I thought, who will it be? Maybe some students. It turns out though, there are a lot of people who are in fact at scale, consuming a lot of compute cycles because they want to get ahead on the quantum algorithms. And one example of that was a couple of researchers out of, I think in case Western or Cleveland Clinic who had an algorithm for fast lookup on MRIs to do tissue matching, basically so you can even discover a tumor while you're doing your first mri. But they couldn't solve it using a classical computer. So they're doing the simulation on Azure. So we're finding, of course the Azure simulator is one. But there are, as you can imagine, other things that you do in order to build a quantity computer that are massively advantageous for anyone who is in the cloud business.

Cool questions.

On that note.

Yes, exactly. If your mind hasn't been a little.

Fried when you're doing blitz scaling or you're doing what you did at Microsoft, you know, the CEO recommends the strategy, but the board approves it. And you know, the board is a collection of people. A lot of them didn't necessarily grow up in a digital age. A lot of them used to looking at backward looking metrics and efficiency. So how do you get a board of directors to be in the right place to sort of make the right decisions around strategies like blitz scaling and when to do it and when not?

Yeah, I can start. I mean, it's something that I've, I would say grown to recognize as one of the foundational jobs of a CEO, quite frankly, growing up even one level removed from a CEO all these years, unlike many of you who are entrepreneurs who deal with boards and investors all the time, I didn't have that skill set. And I realized how important it is when you especially as you said, not everybody on the board is Reid Hoffman or Bill Gates. So we have I think a board size of 13 with lots of different depth in different areas. And the one thing that I realized though is what the board is trying to do is pass pass judgment on your judgment. When we say it's not as much, it's kind of that derivative you've got to get. Right? Because if you think that your job is to explain to the board in full detail the quantum computing, like what causes you to make that judgment, that's going to be hard. It's going to be hard and it's sort of unreasonable even. So, therefore, one of the things that at least I realized is what's the framework for them to evaluate me and my management team on the judgment we are passing. And there, quite honestly, this is where being transparent, being super intellectually honest is the most helpful thing in high risk businesses like ours, where there is so much beta. And when you take a high beta thing and describe it like a sure shot is when you can absolutely lose all credibility. So those are some of the things that I would say, at least I've learned. And the most important thing is that, that don't think that they're passing judgment on the strategy, they're passing judgment on your judgment of strategy.

Yeah, and what Sajay said is exactly right. Two other techniques. One is a board has a number of people. You feel we need to move in this direction. Add someone who's the catalyst to the board, have that be an explicit conversation with them about, look, I think this needs to be going this way. Do you buy in? In which case, as opposed to cause classically the board's saying, okay, is the management doing the right thing? Do they have the right capabilities of the future? This is the second derivative of judgment that Satya is talking about. But it's like, okay, here's someone who is also saying, look, we need to be going here, we need to be understanding that and so forth. So think of it as a team. And then in some cases, I think one of the things that's very useful to do it isn't kind of the full stop or anything else, but to really invest in some board education, you go, this key thing, this battlefield, this product, this piece of technology. They need to understand they're not digital natives, it's fine. They need to understand this and say, look, in order to do this, you guys need to come, we need to set up a one day, two day training thing and you need to come for it right in order to do it. And that's a big cost and a big play. But sometimes that's the only thing that actually work.

First of all, I want to thank you for coming. This has been fantastic. And in the Blitzscaling context, I think context is really important as far as racing uncertainty to where you want to be. We've all seen how Facebook has been in the news with data and some of the challenges around privacy that they've had. And Microsoft I think by and large has been perceived, along with Apple and some others of really taking a kind of a more measured approach to this. But arguably with all the data that Windows collects, with the social graph of GitHub and LinkedIn and some other stuff, you have a lot of data of not just my cat preferences and crazy stuff like that, but real stuff about real companies that matter. And if you really take that to the extreme, like say NATO or something like that, that's using your products and as you're applying AI to this and you're racing in this competitive environment to the future, what is your responsibility as far as privacy and AI and all these issues?

That's a great set of questions. And this is one of those places where I think we are really confronted with scaling while at the same time in that unit of scale building in right in the core, the ability to deal with any unintended consequence first, I'll just say the largest part of our business is foundationally around helping our customers, whether it's individual consumers or large organizations manage their own data. So when somebody moves to Microsoft 365 or O365, we don't start by saying this is our data, it's their data. So tenancy is a massive software construct for us and for most part it's the tenants data. And what is it that we need to do to help them get benefits from their own data. So think of it that way. And our business model helps. When you have primarily a software sort of slash subscription or a consumption based business model, you sort of are incented to do so. And this is not about saying ads are bad or not. I mean, I'm just saying that large part of what we do is foundationally about making sure that customers are dealing with data on their own terms. Now having said that there are three major areas that we are absolutely at work in all of these areas. We have to do a bunch of work. I believe the industry has to do a bunch of work. And then there will be governments involved as well. So one is on privacy. The way at least I approach privacy at this point is assume that this is a basic human right and that it's going to be an expectation of every customer in every context, that there's going to be full transparency and that they will be able to control what data is being collected and what it's being used for. So GDPR is the beginning of that. So in some sense now you even have regulation that you actually have to be compliant with. And we, for example, took the data rights of GDPR and made it globally available across Microsoft. So we are not even saying, because I think it's good for, in fact, the world to not have transactional costs where we fragment on privacy regulation. In fact, if anything, I'm hopeful even in the United States, I know California passed a law which is great, except we want it as a federal law. And the more we harmonize, I think privacy laws, I think we'll be better off. And another area in privacy is Cloud Act. I don't know how much you all track Cloud Act. We, you remember, we took the government of the United States to the Supreme Court twice because of the balance between privacy and public safety. We realize that there's a balance that's required, but it needs to be done within a framework of law. And the Cloud act is very helpful in that. So we finally have a piece of legislation which the U.S. thank God, is leading. Now what we need is the United States to do bilateral agreements between the US and the UK and the rest of the European Union so that we can start establishing a real quorum and an equilibrium around this very important issue of privacy and public safety. A second area is around cyber. And in fact, Reed referenced this. There's a lot one needs to do around cybersecurity. We ourselves are converting our own operational security. All the signal and all the data that we have, whether it's the login data, the security events, the machines we scan or what have you, all that. How do we convert it into better security products for our customers?

Right.

That small business that's being attacked or the account guard is a new product we launched, which, which is a very good example of that, which is email safety. And this is all about converting our operational security posture into better email safety around, say, strontium attacking you or what have you and this is what happened with a couple of senators and that's how we got to this product. Oh, and that's where the digital Geneva Convention comes in. Right. So what we do is good, but we as a tech industry need to come together. And this tech accord a lot of you support and a lot of businesses, I mean a lot of tech companies support it. It's fantastic. This is us taking the pledge that we will use all of our technology to protect the most vulnerable, which is the individual users in countries and small businesses. That's a great step. We need to now even go to the other forums where I think nation states also sign up for that pledge. And then the last one is on AI and ethics and the idea that you should have a set of preferences principles that are even design principles. And we've even talked about it as what is the moral equal or the Hippocratic oath that we as creators of AI can take in order to create AI. That one empowers humans, also protects against bias. So we need to start with a set of principles. But also be open to Reid's earlier point that when sort of there's a technology which is going to have real consequences. Let's take like facial recognition. Having regulation is not a bad thing. In fact, if anything it'll in fact increase the adoption of technology knowing that there's trust in it. And so that's how we're approaching at least our work across privacy, security and ethics around AI. You became CEO of a company massive scale.

You mentioned Growth or Carol Dweck's book as a favorite. You know, of all the books are, I'm sure you talk, talked to many people, but read a lot of books. Any parting book recommendations that have stuff.

Other than Bloodscale, it's like you're going to get me into trouble. He was going to give me that book, the other book that I read and it's kind of an interesting one. And if you think about the foundational challenge we all have, the thing which you got to start is we all won the lottery. Here we are because we work in this industry at a time when this industry and this technology shaping everything and everybody. When I came to the United States in 1988 from India, one of the things that struck me was this country is magical in terms of its, I'll call it egalitarianism. And in fact I've never seen such a large scale population of middle class Americans and the prosperity that was so broad spread, it was just stunning to me. It was kind of like a science fiction place. And in fact, Angus Deakin has this beautiful phrase he uses which he says, the United States is the only country ever in our history to have that blue collar aristocracy. And that's what's lost today. And so there's this book I read recently called Squeezed. I forget the name of the author. It was for me, as someone who's grown up here to really get a better microeconomic understanding how precarious middle class jobs are, middle class lives are and these costs like health care, education, childcare, which all by the way, really technology can help. I mean, the healthcare cost somebody described is not some miracle drug. It's all workflow. And so in some sense that grounding there are so many problems to be solved, so many costs that can be tamed using digital technology. This book and there are many, many I can go on, but it helped me get a better life. And one of the things I'm doing a lot more is reading individual lives to be a lot more in touch with what are the real challenges. And this was one of those very well written books which was helpful for me.

The other thing I'd add in case some folks haven't encountered it yet is Antifragile by Taleb is highly worth reading. So with that. All right, thanks Satya, for coming down.

Thank you so much.