Steve Ballmer 访谈

Well, Steve, first of all, I noticed you prepared some printed materials here for us. Listeners should know we didn't ask for this in any way, but at 10 PM last night, you sent us a PowerPoint deck and said, “I made you some slides. Sorry it got here so late.” Dave and I are looking at each other like, “We didn't ask you to prepare for this. Thank you for the materials.”

Oh, it's just some stuff that I've used with thoughts about how businesses work. I think of this as a time to reflect on things I've learned, primarily at Microsoft, but also with the Clippers, about business. I figured I'd send them to you, and they're PowerPoint slides.

Yeah, I tell you, you mixed a few different opportunities.

Yeah. Always a cheerleader. There you go.

I think the word “cheerleader” is actually in the PowerPoint deck, though.

Yes.

1. Microsoft Loses Its Consumer Muscle

Great. Well, Steve, speaking of reflecting, we sit here today with Microsoft as the most valuable company in the world, at almost $3.5 trillion in market cap. I think everybody would agree it's an enterprise company, and that's largely thanks to you. It is reasonable to call you the founder of Microsoft's enterprise business. That's not a narrative that's often discussed, and we wanted to ask you: How do you feel about the fact that it basically defines the business today?

Yeah, interesting. Very kind. Fathering something, I feel good about that, and I think there's a lot of truth to that. Of course, there are many fathers to the enterprise business at Microsoft, and I feel both good and bad about it, because the truth is Microsoft started out as a consumer company, and we built a very important consumer business. That success translated into the opportunity to go build an enterprise business.

One of my regrets is that we lost the consumer muscle along the way, because I think the ability to be ultra-ultra—I mean, we're a great company. Microsoft is a great company. But to have both of those muscles totally firing, if I'd been able to sustain that consumer muscle—and I had some ideas about why that didn't happen—the enterprise muscle got very big and very strong. I'm very proud of that.

It's also funny when you say “consumer” and “enterprise.” What does it really mean to say “enterprise”? Sometimes it can sound just like backend stuff, and the truth of the matter is that Microsoft Office 365, whatever exactly it's called today, is super important. It was the foundation for having permission to be in the enterprise, and yet it's a product that sits right there in front of users.

So the question is, do you think about users or consumers, and do you think about enterprise? And then there are developers who span both. That's kind of my mental model. Do you have products that appeal to consumers, that IT can handle, and a platform that lets developers build around those, whether they're building for users, users and IT, or, in some instances, just for IT people? There are a lot of tools that are just for IT people.

2. IBM Rules The Computer Industry

Yep. Well, to contextualize all this, we want to go back almost all the way to the beginning, right around the time you joined Microsoft, and talk about Microsoft's relationship with IBM, before the IBM PC and before DOS. Can you catch listeners up who weren't around at that time? What was IBM in that era?

Yeah, I think you called it to us when we were talking to you for research: the sun, the moon, and the stars.

Yeah, I did.

Well, when I got here, the company had started, obviously, and there were IBM computers. There were a couple of others, but literally people would say, “There's IBM and the bunch.” The bunch was Burroughs, UNIVAC, NCR, Control Data, and Honeywell. But they were just the bunch.

IBM did the mainframe, the software, and the service. It did everything in computing—everything. And then you had this little upstart called Digital Equipment.

Yep. Very important in our story, because Dave Cutler, who was kind of the father of NT, Windows NT, came from Digital Equipment Corporation.

They were fighting. They were scrappy. They made minicomputers, so smaller than a room but definitely bigger than a PC, if you will. All the initial Microsoft software was actually developed on DEC computers. Digital Equipment—that's DEC.

DEC had a nice business, but it was a lot smaller than IBM. If IBM breathed, that was the direction the computer industry would go. IBM was the subject of an antitrust lawsuit, shockingly, in 1969 that didn't actually get settled, I think, until shortly after I got here, during the term of Reagan.

And what was the result of that antitrust action? What did they have to do?

I don't remember. It may have been when they had to unbundle the operating system from the mainframe hardware so people could build IBM-compatible mainframes.

3. IBM Comes For An Operating System

One day, shortly after I got here, some guys from IBM called and said, “Hey, can we come see you?” We were going to have to sign an agreement that said we could use nothing they told us, while anything we told them, they could use. These guys showed up and told us, after we signed their agreement, that they wanted to build a PC. They were hoping to get the operating system and some of our language software for it.

And they were coming to you for the language software.

No, they came to us for the operating system.

Ah, now why would you say we weren't in the operating system business?

We had a card called the CP/M SoftCard, or the SoftCard, for the Apple II. It was a card that plugged into an Apple II and ran CP/M, not our operating system. Gary Kildall's Digital Research was the name of the company, but we had licensed CP/M to put on this card that plugged into the Apple II.

Somehow IBM thought they could license CP/M, even though it wasn't our product. They thought they could license it from us.

And we said, “No, no, no, but you can license our language software. But there are these guys down in Pacific Grove, California.” Bill called Gary Kildall and said, “There are some guys who want to talk to you. They’re important.” They went down there, and they didn’t sign the nondisclosure agreement. In the meantime, there was a company here in Seattle called Seattle Computer Products that had a little CP/M clone.

And so the licensing of MS-DOS—which didn’t even exist when IBM approached you about licensing some things—is the single greatest business deal in history. The licensing of that software on our episodes—well, I just think you look at Microsoft’s market cap: $3.5 trillion later. This kickstarted it all pretty good.

There was a company that happened to be here in town. Paul Allen and I went down there, and we met with the founder, who later came to work at Microsoft, a guy named Tim Paterson. We offered him—I think we paid $45,000 or $49,000 for this operating system—because we told IBM, “No, no, we can take care of it.”

There was a kind of famous meeting among Paul, Bill, me, and this guy Kazuhiko Nishi, who ran our affiliate in Japan. We were talking about this, and there were a lot of, let’s just say, four-letter words thrown around. “Screw them.” “Screw them” is five letters, but you get the drift. “Screw them. Screw him. Let’s just go get this operating system. Screw them. We can do this. Let’s go.” That was kind of the theme. Kazo was kind of a cowboy.

Yeah, Nishi is absolutely a cowboy.

So we went and sold it to them for half of what we paid for it, and we thought, “We can do this 10 or 20 times.” Twenty times $21,000—$400,000 against the $50,000 we paid for it. Pretty good deal.

Yeah, it was a little better than that, as you said. Talk us through the structure and how you guys thought about this, because I’m sure you’re right: You did not make a lot of money directly from this deal.

No, we did not. Remember, the key thing was we didn’t charge for the operating system on an ongoing basis. We charged for it 1 time. If you got a new version, we charged another time. We did the same thing for BASIC and everything else because, at the time, you could think we were like a substitute for an R&D department, which means we were fixed-price.

It was only, I don’t know, 4 or 5 years later that we actually switched to licensing per unit as opposed to just a fixed fee. Here it is: Pay us once and we’re done.

But the ultimate thing that you guys negotiated was a nonexclusive deal. You could sell this operating system and your language interpreters, but they also said, “Look, instead of us building everything all custom, we want to use some industry-standard parts and components, because that’ll let us be more agile,” et cetera. So they didn’t come in opposed to any of this.

They knew that was our business. They knew that was Digital Research’s business. They wanted to use an Intel part versus their own proprietary part. They didn’t ask Intel to make them a custom part, either. The notion was, “We’ll move fast. We’ll get away from the IBM bureaucracy by taking this approach.” So I wouldn’t say that was the hardest convincing, if you will, in the story.

But what ended up happening after all these years—and I imagine it only took a few years to see it play out—was IBM sold a ton of IBM PCs, and DOS was the operating system. Then everybody else adopted DOS because all the application makers, all the software vendors, were targeting DOS as the platform, and so Microsoft sort of accrued a huge amount of benefit. You became the point of integration. In the old world, IBM would have accrued that sort of platform benefit. Did they see that they were selling a lot of computers and making a profit?

Also, they would have been making more profit than we were at the time, just the way pricing worked. There was a little twist in here, though, I should throw at you, if you’re curious.

These things had something called the BIOS—basic input/output system—which was the lowest, lowest layer of firmware, sort of first-level software built into the hardware. IBM had its own BIOS, and some applications became BIOS-dependent. So then the question was, who was going to do an IBM-compatible BIOS?

We weren’t going to get into that game. We didn’t want to have that intellectual property; there were other arguments. But there were people then—Compaq was important and ultimately became the big company. I don’t remember whether they wrote their own compatible BIOS, but they were the first to be IBM-compatible. There were plenty of people who ran MS-DOS who were actually not IBM-compatible because they didn’t do a compatible BIOS.

I see. So IBM sort of thought, “Oh, we’ve got some protection from Microsoft kind of disintermediating us from all the developers and all the potential customers, because targeting our BIOS is going to be important and unreplicable.”

Now, the one thing you have to remember, because we live in the modern world now, is that when you say “all the developers,” that wasn’t a long list. Remember, there was no software industry to speak of when we got to the creation of the software industry. There were a couple of software companies that made packages for IBM mainframes, but almost everything was custom.

So really, I would say we and a few other companies—but I’m going to say we—defined what a modern software business looked like. The notion that there could be lots of developers—and, yeah, there were some—but it’s not like we think today: “Oh, there were developers doing lots of standard applications.” No. There was no licensing model, no business model, no nothing. VisiCalc was around.

So it would have been counterintuitive, or required too many mental hops, to think, “We’re IBM. Wait, are we giving away the future by allowing someone to distribute a widely available operating system that ends up being the target everyone standardizes on, which eventually created all of modern Microsoft?”

Exactly. You sort of can’t blame them because there was nothing to build off of. But, yeah, one of the things my little PowerPoint here says is, “Luck is important in the creation of great companies.”

It is.

A lot of people sort of say we’re masters of the universe. We figure everything out. We never have any luck. It’s because we’re so talented. Sure, there are talented and hardworking people. Most people have a little luck in their story. This was our big luck, clearly.

But when you were negotiating this, signing it, and then during those first couple of years before the clone market really took off, did you think that this could happen?

No. I can’t remember what year it would have been, but Andy Grove, who was running Intel at the time, said, “Yeah, pretty soon we’ll be selling 100 million PCs a year.” I don’t know, sometime in the ’80s, I think. It might have even been in the ’90s.

Bill and I laughed and said, “Ah, that’s not going to happen.” We invested big-time, and if it did happen, we said, “That’s great. We’re not going to underinvest.” But we thought, “Ah, he’s crazy. This market will never grow like that.”

I would say we classically underforecast. That was kind of our tendency.

So the deal gets signed with IBM, you end up shipping DOS, it goes on the IBM PC, and it’s selling like gangbusters. When did you start to realize, “Whoa, what we have here is actually leverage over the ecosystem? We actually are becoming the important layer that ties this whole computing world together with the operating system, personal-computing layer?”

4. Windows Breaks Free From IBM

Well, I think by the mid-to-late ’80s. You make it sound very strong. No, we didn’t feel very strong. There was IBM, man. IBM was still the sun, the moon, and the stars. That didn’t change.

I would say we didn’t drop that theory until well into the 2000s. Into the 2000s, Lotus Notes was coming for us, and that was the mid-’90s and beyond. But maybe you could say later; we weren’t an enterprise company. If you looked at the enterprise, the enterprise was still IBM.

We used to say we had to hang on to IBM, because if we ever let go, they might trample us. We called them the bear, and the bear—you had to stay on. Of course, the graphical user interface was kind of coming out of Xerox PARC at the time, and Apple was doing its thing, and we started thinking, “That’s another disruption that could blow everything up.”

So I would say there was no sense of confidence about controlling the ecosystem well into the ’90s, before I think any of that—or at least for me.

When did you start to feel like, “We’re getting out from under the thumb of IBM,” and maybe walk us through a little bit of the OS/2–Windows world?

We’d been staying with IBM. They decided they wanted to build something that was sort of their operating system and sort of not. This was 1982–83. We and they would collectively build part of it. We would be able to license it to others. They would build a value-add layer that was a database and an emulator.

Crazy to say now, but we were going to work on the operating system and what was called Presentation Manager. Call that the graphical user interface. They were going to have rights equivalent to ownership in the code we wrote.

This sounds so convoluted. It was so convoluted, man. There was a time when I made 16 trips to the East Coast in 16 weeks, most of them to South Florida, a couple of them to New York. I’d leave on the red-eye, the Delta Dash flight, at around 11:00, get into Atlanta around 5:00, get the flight to West Palm Beach at about 7:00, get in and be able to be at a meeting at 9:00 at IBM, and then work all day, catch the 7:00 flight home, be here about 10:30 or 11:00—24 hours down and back—because if you’re building something together, remember, there was no real email at the time, right?

We were literally shipping disks back and forth. Then they decided they were going to do the Presentation Manager piece in England. So there were also a lot of flights to England.

Then Texas is where the database and the communications were.

IBM—this sounds like Boeing.

Yeah. We called it the joint development agreement. It was the price of staying involved with IBM, and it was convoluted. For speed of action, we kept going on Windows, which we had started.

For the listeners, everything we're talking about is the OS/2 operating system. Basically, there was OS/2 Extended Edition or something, which had their edition, and Windows was like your plan B. It was like your side project.

No, Windows was our plan. They wanted to do this new operating system, and we convinced them, “You've got to have a graphical user interface.” We tried to sell them Windows, and they were resisting.

Okay. So it almost seems like you're humoring IBM at this point with, “Yeah, let's do OS/2 together. We really think the future is Windows.”

“Humor” is more than I would say. My job was managing, by then, system software. So I had Windows. I had shipped it when I'd been the development manager for Windows 1.0.

The great videos of you from the Windows 1.0 launch.

But that's the sales side. I actually managed the engineers because the guy who was doing it wasn't being successful, and we had to ship the thing. That's when I learned something about engineering management. The engineers basically had to teach me to be effective.

We were trying to keep up with OS/2. Bill was very frustrated with IBM. I was frustrated, but I knew my job was to ride the bear. Bill was pushing Windows hard, but we still suspected OS/2 could be the winner because it came from IBM. But we couldn't just stop for 3 or 4 years. We couldn't make the mistake we sort of made in the thing that became Vista.

So we kept going with Windows, and we kept going with OS/2. Then, in May 1990, they came along and shot us. I was out running with my wife.

IBM shot you.

Yeah, they divorced us. They threw us out.

I thought the story was that Windows was gathering strength, and you all thought, “Maybe we can step out from being the little brother.” They came after you.

No, no, no, no. They had a new leader by then, a guy named Jim Canavino. He was getting frustrated with us because we were still selling Windows and promoting Windows.

I mean, look, this was our first antitrust problem. I don't know if you guys know this, but the FTC at the time thought we and IBM were working to divide the market because we had done some positioning: What's Windows good for? What's OS/2 good for? We and IBM had done that, and then they said, “No, you guys are colluding.” That's when we first got the attention of antitrust authorities.

This is even before the per-processor licensing issue.

Yeah, that came later. That came with the DOJ. This was an FTC case, and they started it in basically 1990—just as we were getting, I think 1990, maybe 1989, as we were getting our divorce.

My wife and I were remodeling our house. We were living in a condo. We stopped on a run, used a restroom or something. I picked up the Wall Street Journal, and I read that IBM was divorcing us.

I picked up the Wall Street Journal, and I read that IBM was divorcing us. So what does that mean? Walking away from the OS/2 collaboration?

Huh?

Basically, they kicked you out. They kicked Microsoft out and said, “We're taking OS/2 in-house.”

Exactly. Exactly.

And so you're sitting there. Windows isn't powerful yet. Windows is this fledgling idea. You still had something called the 640K barrier. You couldn't speak to more than 640K of memory. You didn't break the 640K barrier until, I think, Windows 3.1, which I want to say was 1991 or 1992.

So you're on this run, and you see that IBM is divorcing you. You don't really have confidence in Windows yet. What are you feeling? What do you think is possible, Mr. Wizard?

Whoa. Shoot. Oh my God. We were so, you could say, energized, if you like. Scared also works. It's like, “Oh my God, now we have to confront the bear.”

You're already a billion-dollar business at this point. By the end of 1992, you were at $2.8 billion in revenue. IBM is still IBM, but you're still a pipsqueak to IBM.

We're still a pipsqueak to IBM. And remember, we had no enterprise presence, and IBM had a totally dominant enterprise presence.

So who's using Windows, and how are you selling to them at this point?

Interesting. Single copies, some hobbyists and end users—somebody who says, “Hey, I really want to use a spreadsheet”—and a lot of users in enterprises. It wasn't going through IT.

You'd have a user who would buy a PC on the expense account, probably for the department, buy a copy of Windows, buy a copy of Excel at an Egghead Software—it was a software retailer at the time—and bring them in and use them. Then IT started to get nervous about that.

What the hell? IBM's going to stomp us like a bug. You just took as a given assumption that if IBM wanted to stamp this out, it was going to happen.

Oh, yeah. If we wanted a future, we had to play with them. That's why we were, quote, “riding the bear” the whole time, because they'd stomp us out. They divorced us in 1990, and then we said, “Oh my God.”

Okay. So at this point, your business, even though it's a billion-plus-dollar business, is selling to retailers to sell copies of software: DOS, Windows, languages, and apps. DOS was always sold to OEMs.

Yeah. Not always, but so much the lion's share. It's worth saying it was only sold because you needed a BIOS. Remember, you needed a BIOS, so you had to have the hardware vendor build the BIOS into the machine, basically.

So you've got the OEM business, which was the biggest part of the business.

Yeah, and then we had this retail business. There was no notion of enterprise licensing.

You've got no CIO relationships, no enterprise agreement, no—

We had a couple of CIO relationships.

The Air Force was the first big Windows customer.

Your first big Windows customer, at least as I remember it, was the U.S. Air Force, and they were buying single copies of Windows.

When you say government, there's really 2 governments in this country. There's government, and there's the military. The military is a very disciplined, advanced user of it. They're just better. They're more professionally run than most parts of government. So, yeah, it was the Air Force.

So you got, like, a little bit—

Yeah, we had 1 or 2 customers just to prove we could actually serve big customers.

5. Ballmer Builds Enterprise Muscle

As we understand it, you had this realization at this point: “Well, I'm going to go figure out how to do what IBM does,” like you personally. To put a finer point on it, the thing that we said in our episode—and I'm curious if it's true or not—is this was not Bill's passion area, and you sort of raised your hand and said, “I'll go figure out enterprise sales.”

Oh, yeah. No, no, that's for sure true. Bill's passion—look, Bill had passions in a lot of places, but you'd say the apps group and what Windows could deliver to the apps. Quite appropriately, I'd say that's where a lot of Bill's brain cycles went. I had also hired Dave Cutler. Dave Cutler had been the architect of the VMS operating system for Digital Equipment Corporation.

And, you know, we had DOS and Windows. When we were talking to Cutler about coming here, he said, “I don’t want to work on any toy operating systems.” I had to say to Dave, “Good thing, because we have a toy operating system.”

But Dave is the key to getting us there. We said, “Look, you’ve got to build an operating system whose API looks like Windows and whose user interface looks like Windows, so developers can be familiar with it and write apps for it.” You might make some changes because you have to, but it’s got to be a robust operating system. It’s got to have a secure kernel. It’s got to have all of these things.

The product set that you had wasn’t really enterprise-grade yet.

No, we had a joint development agreement—a joint agreement on LAN Manager—with a company called 3Com. It wasn’t all our stuff. We had a development agreement with a company called Sybase to do the SQL database, because we were trying to figure out all these pieces that IBM would have. We didn’t have any of that. An operating system alone is not going to do it. You need all these other components.

If you want to have back-end infrastructure, we started scrambling on that in the ’80s. We had all these infrastructure pieces that we had to build if we wanted to sell to, I’ll say, business customers. We weren’t even thinking about—when you say enterprises, sometimes people think very large companies—but we couldn’t sell to companies of 20 people without some of this stuff, or 50 people.

You talk a lot now about this management concept of building muscle. Is this where this came from—that you should always be, as you use the phrase, in the weight room building muscle ahead of what you need? Were you and Bill thinking this way in the ’80s, like, “Hey, we need to be building up this muscle across all parts of computing and business computing”?

Well, Paul Allen—I mean, Paul’s the key. Paul is the one who said, “We’re never going to be a hardware company.” When the Altair came out, the first real microprocessor-based computer, Paul said, “Okay, let’s write all the software that these things will ever need.” Bill and I had a lot of the execution around that, but that was the push.

Paul was cracking on me in the early ’80s to start building an apps group: “Come on, Steve, come on, Steve. It’s not just systems. We need to have applications also. Any code that executes on a microprocessor, we should have a player in that market. There was a VisiCalc spreadsheet. Come on, the word processor. Come on, come on, come on. Let’s get the talent. Let’s get going.”

We were doing mostly college hiring at the time, so, okay. Then we met this guy, Simonyi, who had been at Xerox PARC.

Charles Simonyi, right?

Charles Simonyi, exactly. We met him through a mutual friend at 3Com Corporation who had been at Xerox PARC. He really was the first leader of the apps business.

But we licensed—I mean, look, we worked with other people the way IBM worked with us, right? We went to Sybase and 3Com and said, “Let’s work together.” It wasn’t exactly a JDA, a joint development agreement, but we worked with those guys the way IBM worked with us. The analogy now is a little bit Microsoft working with OpenAI. When the big company works with the new company, how does that all play out over time?

But I took over systems software in ’84. That’s when we were starting all this stuff, and you could say I was a little bit more enterprisey.

Yeah, I’m looking at your chart here that you made for us. You’ve got 1992 to 1998 titled “Liftoff,” and that’s after the era where you talk about “Enterprise Start.” You have your role switching from the OS division in the previous era to sales. The liftoff there, though, is mostly on Windows and applications. The liftoff isn’t really enterprise.

I mean, look, it was not until the late 2000s. People would say, “You guys might find this funny, or maybe you even know it. Customers say, ‘You’re not an enterprise company. You’re not an enterprise company.’”

As late as when?

Oh, late 2000s. Absolutely.

You’re not enterprise-grade. You’re not enterprise-ready.

I heard that so much.

Who is?

Yeah, you had Oracle out there. Remember, there were still mainframes and minicomputers, and people—you know, those things were enterprise-ready. IBM still had products. You didn’t have enterprise support. Our licensing had to evolve in the early ’90s and then again in the late ’90s. No, we didn’t have those things. So, no, we weren’t an enterprise software company.

In the 2000s, certainly it wasn’t before 2005. It wasn’t at the beginning of my tenure. We were still trying to prove that we were an enterprise company. Now I just find it cuckoo that all Microsoft is characterized as an enterprise company, which—I think it’s more complicated than that, but I’m not going to say that that’s not the primary muscle. For sure it is. But I—the company, I mean, I was hell-bent and determined to prove we were an enterprise company.

Why was that? Why did you feel like this—let’s call it 1992, 1993, 1994—why did you feel like it was so important for us to attack that market?

Easy. Because that’s where IBM could pound us, squish us like a bug. If we couldn’t sell our stuff to businesses, only to consumers, we knew that by then we’d only get so far, because enterprises wanted some features. Enterprises don’t like, “Okay, you can go to ComputerLand and buy a few copies.”

The consumer market—I mean, we’re pre-mobile, right? So, pre-mobile, the consumer market, pre-internet, is big, but it’s nowhere near IBM’s market in the enterprise market by revenue.

No, for sure not.

So, we’ve talked a lot about the products. Let’s talk about the go-to-market motion and this invention of the Enterprise Agreement. What are the key pillars that you came up with for the Enterprise Agreement, and why did they exist?

Okay, our first software pricing and packaging model for the enterprise was not the Enterprise Agreement. First, we sold you discs. Second, we came up with this notion of what we called Select licensing, and you could make your own copies and just report how many copies you sold and pay us what you did.

That sounds rife with challenges. You tell us how many copies, and just pay us what you did—the enterprise honor system. Astonishing.

That was of Windows and Office. Windows typically came with the hardware, so you were mostly using the OEM channel for Windows.

Yeah.

Even to this day, upgrades and stuff are sold direct to enterprises, but a basic computer that comes to an enterprise would have the operating system licensed to the OEM.

So we were on—you can call it the honor system—but we just couldn’t make people buy discs from us, or CDs. Enterprises didn’t like that. So we had this thing called Select, and Select had 2 problems. Number 1, it was very hard to count copies of the software you deployed. Number 2, we were selling upgrades and new licenses, and upgrades were less than half the price of new licenses.

What does that mean? The company was headed to a world where its revenue was half of its existing revenue.

Yeah, unless you’re growing new customers, new logos. Phenomenal clip.

So it was a real problem. Bill and I had always dreamed of this thing where you get some recurring revenue. Then we said, “Okay, well, why don’t we just do a license where you didn’t have to count the number of licenses you printed, just the number of computers? It made life simpler.”

We said, “Instead of selling you a new license and then God knows when we would sell you another upgrade or whatever, we’ll do something that just says, ‘Hey, look, you sign up for 3 years, you pay us per machine, and you just pay us the same amount of money each year for 3 years.’” It sort of let us jimmy up the price of the upgrade.

You said you had a pricing problem, and we solved the difficulty-of-administration problem.

And that was the Enterprise Agreement.

And was it from the beginning that you got everything?

No, that was a special Enterprise Agreement. You got all the upgrades during that 3-year period to the products you licensed, but you were still picking and choosing: “Oh, I want Excel. Oh, I want…” We were encouraging you to buy Office.

Yeah.

But we also had this all-you-can-eat license. I can’t remember what we called that, but basically, I think you counted the number of employees and you could use any of our software for anybody.

So we just tried to go simpler and simpler and simpler in the administration, with recurring revenue that didn’t decline over time and as much as you wanted to eat—the upgrades, everything. We did want, essentially, what you have now, which is a recurring services business, but we didn’t have the cloud. We weren’t delivering things, but we were already on that path.

We started with Energizer. You guys mentioned what we did with Energizer, which is where we wanted to run their IT department, right? They were the pilot customer for this concept. They were the first customer. I talked them into it. This is beyond the Enterprise Agreement. This is where we actually wanted to run their stuff, because we did want to get to this recurring-revenue thing.

David was referring to this concept earlier. We talked about it a lot in our Microsoft episode and then on our Epic episode: this sort of genius idea of, “You will get included in your license a whole bunch of software, even if you’re not ready to use it yet.”

If at any point you’re considering buying a different software package from another vendor who just makes this one thing, they look in their paperwork and they’re like, “Oh, wait. Actually, we get that from Microsoft for free as a part of the thing we’re already doing. Let’s just do that.”

As long as you’re developing a lot of software every year, you can indefinitely just make more and more and more stuff so that your customers don’t need to look elsewhere as they expand their software needs.

How did that come about? Let’s start with Office.

When we created Office, Bill really drove that integration. We had PowerPoint, and then we put these things together, and people would complain. We didn’t always sell Office. People would say, “Our customers—our users—don’t use Excel, so we don’t want Excel included.” Okay, we had a licensing option for you. But it became easier and easier.

Departments were always running it at the time, and still do to this day. So we did sell you things that you might not be using. But also, if you’re trying to deploy it across departments, we already had it all for you. You may want something different for this department, but we had it all for you. That was an attractive thing for people.

There’s an insurance aspect that I learned IT people really want. They want peace of mind. That’s part of what it means to be an enterprise: “I want to make sure everything’s secure. I want to make sure everything is well managed. I want to make sure everything is well paid for. I want to make sure there’s somebody to call if things go wrong. I want to make sure I bought everything. I don’t want to look bad because either I paid too much or I have holes in what I bought for people.”

I view this—and I probably evolved my view over time—this way: when you sell to the enterprise, you have to provide peace of mind, which is kind of like an insurance policy. Buying more than you might be using, or more than some users are using, is an insurance policy. Software has zero marginal cost and zero distribution cost, so we were happy to mail you a few more discs if you needed them. But we weren’t even mailing discs by then, because we had the Enterprise Agreement in place.

At a certain point along the way, you get to—I want to say the holy trinity, but I think there are more than 3 pieces of this—the real killer suite in enterprises: Windows, Windows Server, Active Directory, Exchange, Office. All of these pieces of software work in orchestration to run your enterprise.

Your users do their email on Outlook, which is part of Office, which runs on Windows, which uses Exchange, which uses Active Directory and SQL Server—all these things. How long did it take to get to that point, and what went into it? To my mind, that’s when the enterprise is firing on all cylinders.

That really comes with the email boom, and the email boom is the late 1990s through the beginning of the 2000s, because email is sort of the cart that pulled the whole—

No, it’s the locomotive. Enterprises wanted email.

When Accenture became a company, we started a joint venture called Avanade to help do essentially the holy trinity—to help install it. We needed support infrastructure and partners who knew how to set up the servers, provision email, and put all that in. We needed partners, and we didn’t have enough capacity. That’s why we started Avanade with Accenture, which is a big company at this stage, and that was in the 2000s. I went on the board of Accenture.

All this to say, the way you could pitch an enterprise is, rather than any of these other value propositions—David listed off a whole bunch of software—you could say, “You guys want some email, right? We have the most reliable, robust way for your enterprise to adopt email, and it’s going to come with all this other great stuff.”

Everything was nicely integrated because, remember, you needed Active Directory to manage file shares, to manage printers—I mean, it was used for a lot of different things. So it really did all come together as part of the integrated proposition.

Like you say, you guys sort of made fun of the notion that we called all that stuff the back office, as if that was the Ministry of the Interior.

Oh, so wrong. So wrong about that.

He took that as a signal that Bill just didn’t care about this.

Oh, completely not right. I wanted to call it the back office because you needed to buy the Office and the back office. The user, the consumer, saw the Office, and the back office was the thing that was in the server rooms and data centers, although a lot of them were server rooms. It’s the same thing these days, but cloudized.

All right. So, as we were preparing for this, there were a bunch of big questions that we desperately wanted your take on. A big one is around one of your most iconic moments: 1999, the “Developers, Developers, Developers” speech. I’ve probably watched this clip 20 or 30 times. Almost everyone listening has seen this clip. What is missing from this clip is all the context around Microsoft and what’s going on in the world at this time, and what you need to accomplish as a leader of this company. Help us set that stage, and then help us understand why you went on stage that way.

6. Developers Keep Microsoft Competitive

Remember, by this time, we’re not through with our IBM competition, and we’ve got Linux competition on the docket because Linux is competing with Windows Server. Linux is competing with Windows, and there’s a thing called OpenOffice—open-source software for Office—that’s competing with Office. So we have all these things going on. We haven’t beaten Lotus Notes yet, and we’ve got antitrust issues, of course, by then.

The culmination of the DOJ suit is happening within 12 months of this moment, correct?

But it’s clear in all these competitions that the thing you need is third parties that reinforce what you’ve got, adding value around what you’ve got. I could say, “Run on your platform,” but I’ll come to that later if you want to—what a platform is and isn’t. If you want to do that, it’s kind of interesting, particularly since everything’s called a platform these days.

Let’s take an aside here. Give us your definition of a platform.

You could call it anything that is extensible, and it’s the extensibility that, quote, makes it a platform, because you’re going to get people to extend the value you add.

The question is—and the reason that’s important is—that applications are platforms too, not just developer platforms. When people say “platform,” they might mean Azure, AWS, or, in the old days, Windows, Windows Server, Unix, then Linux. Yes, those are platforms. You extend them, but you also extend Office. You add value. Partners plug in, they write applications, they use the file formats. All of this stuff is a platform.

Part of the issue for Microsoft, I think, is that if you see yourself as just a platform company, platforms need apps. You want to have the top first-party app that runs on your platform. Otherwise, your platform can’t get good. Office was the best first-party app on Windows, and that’s how things get good. Outlook was the best first-party app on Exchange. There were other clients at one point, by the way.

You really do want extensibility in your apps in addition to your platform. You want to make sure you own a first-party app in addition to the platform. I think you can get stuck in the mud if you say, “We’re just a platform company.”

I think we got it into our corporate mindset that we were a platform company far more than I ever intended. There were people telling me in the mid-to-late 2000s, “Well, we can’t do that. We’re a platform company.” I said, “Yes, we can do that.” By 2010, I was just frustrated with myself and my inability to get people out of the “we’re just a platform company” mindset.

To this day, you have to think app with platform. You have to think extensibility of the app and the platform. I think we got caught on that. Maybe I got caught on it for a while, and I certainly got caught in my inability to tell people what the company needed to do, because people had such a culture then of saying, “We’re a platform company. We’re a platform.”

So now I go back to, “Yeah, developers, developers, developers.” I’m trying to tell people at that time that third parties really mattered. You got different opinions inside Microsoft.

What event was this at? A developer conference, I think.

So it’s for external developers.

External developers. And you know who Windows’ number-one client is? Is it Office, or is it all developers?

You ask the Windows team, it’s all developers. You ask the Office team, “Come on, you’ve got to do for us what we need you to do.”

You have to be able to communicate that you really care about developers who are not your own, that you really want these things, because they may think, “Oh, it’s all about running Microsoft Office.” We just had to tell people, “We want you. We want you. We want you. We want you.”

I think we got caught in thinking it’s all about third parties and not also about our first-party apps. That’s why you ask, are you—the word “consumer” sounds unserious—are you for users and for enterprises, which really means IT departments? Or are you for users and not IT departments? Do you allow all aspects of what you do to be extended by developers? That’s the frame I believe in.

We had some issues over the course of where we went in the 2000s. We can talk about that if you want to, but go back to 1999. Come on, we need you. We need you guys on Windows. IBM’s still selling OS/2. Linux is right there on the horizon. It’s coming like a freight train.

Is the web starting to enter your psyche at all? The web’s part of that, right?

We’re trying to get people to write for Windows Server.

Good point.

We’re trying to get them to extend ActiveX controls. I think this is part of Netscape, right? We’re part of the browser. We were trying to get our browser to be a platform—a unique platform. “Embrace and extend,” I think, is what we said. We’ll embrace the internet, and we’ll extend with these ActiveX controls.

We need developers to do ActiveX. We need them to do Windows Server. We’re just sort of getting ready on .NET. I have my own wild style, and really, how do you end a speech? You tell people you love them, that you want them. That’s sort of the call to action. And that’s where I think the “Developers, Developers, Developers” thing came.

I mean, before that one, there was a different video that people sort of characterized as “I love this company.” No, there was my Windows video. I don't know if you've ever seen it.

Oh, of course. But wasn't that a parody? Don't people misunderstand?

It was for fun. It was just a fun thing. It was not a real speech, and it was for internal consumption.

Where you're saying, “Yeah, it was for sales for this low, low price.”

Yeah. I mean, there are a lot of little nuances in there. We were trying to get our people pumped up about Windows.

What I was looking for there is—the “developers, developers, developers” speech is one where you feel like we haven't really won the last battle yet. We're still in this death grip for enterprise developers, or this death fight against IBM. And yet there's now Linux and the web for these more independent or platform-of-the-future-looking developers. In some ways, we're desperate to sell, to win, to say, “Hey, we have a great platform here. You need to come use our stuff.”

Exactly.

I can't remember whether we're pre-LAMP or LAMP by then, but I don't remember. There's some infrastructure on top of Linux that people are using to write, let's say, their backends, not their user-facing code. And we had tons of competition.

The interesting thing is, people say, “Only think about your customer. Never think about your competitor.” I actually think you have to think about both. Ironically, we were pretty consumed with our competitor, which I think was essential, and we were pretty consumed with doing new things. But the competitor thing wound up being very important.

I mean, we have no business. We're not in the enterprise. We could lose Windows on the client. We have to, you know. And the company—we weren't really self-confident. The DOJ was really self-confident that we were kind of a lock and there was no competition, and life was easy. That's not where our heads were.

7. Windows Everywhere Goes Wrong

There was some time in the 2000s when I think we did think that extending—we did a slide once called “Windows Everywhere.” We used to use this on all these devices, and we became too wedded to extending what we had versus jumping to something new because, in a sense, we were too confident.

We were too confident that if we only Windows-ized something, it would work. You guys make a point in your episode on us. You guys call it sticking with Windows too long, but I don't think we stuck with Windows too long. I think what we did is we tried to put Windows in places that it didn't naturally go, and we tried to be too Windows-like, both in the API and the UI, in some things.

Mobile being an obvious example—Windows Mobile.

Exactly. And the car.

And the car.

We did a layer on Windows that, when you hooked your PC up to the TV, had a simplified user interface for this.

It wasn't just Media Center, right?

It was some Media Center.

Okay. Media Center.

Yeah, Media Center.

Exactly, right.

So we became convinced, either out of some degree of paranoia and some degree of confidence, that our birthright here comes from Windows. That's our permission to enter the area. But then, in some areas, it just wasn't going to be extensible. So there was both a fear and an overstated confidence in trying to take Windows everywhere.

Let's jump to this point, but what is the generalizable lesson here? You have Windows, this amazing piece of software with this tremendous multisided network effect around it. The logical thing to do is to continue to try and extend it and say, “Geez, wouldn't it be nice if the next great technology wave was also Windows and that worked for us on Windows Server?”

So it's not like we didn't have an existence proof that the thing could work. But, you know, if you're going to—in my little deck I gave you—

Yes, please.

You know, if you're trying to skate to where the puck is, if you're trying to recognize—what did I call this?—capabilities, you know—

If you're a startup in something, there's an ongoing business. You just figure, keep enhancing your products. There's a line extension: “Okay, we're going to add networking to Windows.” No problem. You still call it Windows. It's related but new.

SQL Server, for example, was that for a while. It was related because we had a back-end platform. Dynamics was somewhat related—our accounting, et cetera, stuff—because there was some enterprisey sales, but it was really new. And it turned out the phone was more like a startup.

The phone was more like a startup in recognizing and thinking about things and then asking yourself, “What capabilities do you need?” I say, “Get in the weight room.” You have to develop capability.

Take a look at a capability we developed that is now essential. We didn't build it for this reason: hardware design. Microsoft's a major hardware design company now. I started it out mostly on the client side to help client-side devices—Surface, Xbox, Surface Phone—and guess what? They use that mostly now in Azure data centers.

I think the guy who actually runs hardware design used to be on Xbox. The back-end hardware design for the data center, the chip, et cetera—infrastructure. I'm pretty sure there was a lot of talent we brought in. So building capabilities is important.

We built some capability, but we didn't build enough capability. We didn't see things as different enough. We thought, “Okay, let's try to keep the comfortable Windows user interface because people understand it.” It wasn't right for the phone. I don't even remember what processors we started out on, but I'm pretty sure we started out on Intel. Of course, that wasn't right.

We tried to keep too much consistency, both out of a fear that this was our permission to exist and out of a self-confidence that we had to put Windows everywhere.

So when should a company that has an existing, fantastic business say, “No, no, no. We cannot extend our existing franchise to this new world. This new world is going to be dominated by some new paradigm where we have no advantage”? How do you play that? Then do we choose to get in?

Exactly. Then you have to choose to get in. I would say 2 things were true at the time for us. This is specifically about mobile. It's also about something else. It's a little bit about search, too.

There were 2 things that are true. Number 1, you have to be focusing consciously on the issue. It's easy to get caught up in what you have. You get caught up in what you know. You get caught up in the capabilities. That's why I say to myself, you explicitly have to think about it.

And look, if we hadn't developed a bunch of the capabilities we had—communications, AI—if we hadn't built Bing, the company wouldn't have had capabilities.

I was going to get into—yeah, you built some capabilities in online services that we'll come back to.

We built some important capabilities, but we didn't realize the businesses were different enough to harness those in the new ways. I'm proud of the capabilities we built. We didn't apply them the way we should have.

Where did we learn to build Internet-scale infrastructure? Some with Azure, some even more than Azure.

No, even more than Azure to get started: Office 365, what's now Microsoft 365, the Office back end. That got critical mass as a cloud infrastructure before Azure did, and even more so with Bing. So we developed the capabilities, but then you look at the product, and what was our strategy for Bing?

There was too much based upon Windows integration. You have to say, “This is separate.” Before the Bing rebrand, it was like Windows Live, right? Windows Live Search, right? Everything was Windows Live—OneDrive with Windows Live. But the file sharing—I mean, look, Google's done the same thing. And you have to ask, where do you run out of gas? Yeah.

Yeah, because you could make the counterargument: Google is running away with the market. It's very good technology. They've perfected the user experience. They have scale, and you need scale in this business. It's a runaway train that we're never going to catch. Thank God we have Windows to be able to have some way we can attack them from the side. And with Windows integration, maybe that gives us a fighting chance.

That didn't end up being true, but you can paint that narrative, at least: We can't fight Google. Tell yourself that. Look, how late were we to search? The answer is: When did Google start? '98? '98. Okay. And we jumped in in 2003. I think we pushed.

Now, you'd say 5 years is a lot, or you could say 5 years isn't that much. You could say we had no birthright. It was just a completely separate thing. We had no capability. We had nobody who'd grown up in that world. We had some research guys in Microsoft Research who could start getting us there. We took talent that was doing other things at Microsoft. It's hard to go get new talent because search is brand-new. There were people from Inktomi; Google had sucked them up. So it took us a while to get off the ground.

It took us a while, even—to be fair, I think this is something both Bill and I debated, not just with each other, but we kicked around too much how important, quote, the verticals in online services would be versus search. And portal is generic. Search and portal is generic, but remember, we had a thing called Expedia. We built a travel site. We built a local information site called Sidewalk. We had a car-shopping site. What did we call that thing? CarPoint.

How much would the verticals be worth? There was 1 vertical that mattered, except it wasn't really vertical. It's called all shopping. There was all information and all shopping. And you get to doing all these detailed, specific things. Remember, we did a portal, and then eventually we did search a few years later. No, we were just off. We had the wrong thing stack-ranked in the wrong way, in my opinion, with 20/20 hindsight. We were spread too thin.

So you said, when should you get into a new thing? Well, you probably shouldn't get into 5 new things if you really only have the talent for 1 or 2 new things. That's number 1. Scott McNealy used to have this expression: We've got to get all our wood behind 1 arrow.

It's nice to try. I was listening to you guys talk about Amazon and how they were going to try small things, but they also put in a small cost structure. We put in a big cost structure because we were already all in when we got into something. In this particular case, a few years later, what do you do? You get stuck.

Well, we have permission to come from behind in a certain way here because we've got Windows. It's your point. Exactly your point. There are lessons to be learned, but for a company that's got an established business, being able to get all the way outside of yourself and say, “Is this really like what we're doing?” Because you really want it to be. You really want it to be.

Or does this really require a different approach that doesn't totally ignore, but doesn't take into account what you own any more than the person starting it? Can you hire new capability, or how do you build new capability? If it's not like what you're already doing, it must require new capability. If it's exactly like what you're doing, then you'd be doing it, and you should be great at it. You'd be great at it.

Just look: 2 models worked in phone. Build the hardware, capture the profit, and have a back-end monetization system that even lets you pay the phone manufacturer. That worked. Android/Google. So 2 things worked. That's it. We weren't in either one. We needed new capability. We needed a new idea. We couldn't use the Windows user interface. There were a bunch of things, but you have to go all the way.

And yet we had a Windows Everywhere slide. It was on the slide. I don't understand why it didn't work. You get locked in your model. “We're a platform company.” No, we're an app-and-platform company.

On our episode, we threw out the idea that Microsoft's competitor—the truest form of the competitor it should have taken on in mobile—was not actually Apple. The iPhone is not the bogey. It's a pretty different thing. At that point, you were not a hardware company. The bogey was Android.

They were monetizing a different way, through advertising and through giving it away for free. Microsoft always monetized through licensing revenue. It seems like until Android took off, Microsoft actually did have an opening.

Okay, that's the second Christmas. What year Christmas was this? There was a Christmas—blah-blah-blah year—and it was about being on time with the stuff we needed for Verizon. There was a Verizon design win because Verizon, by now, is really feeling like it's getting its ass kicked. The iPhone launches on AT&T in July 2007, and it might have been Christmas 2008. Yeah, because the App Store—Christmas 2008. Possibly even 2009, but I think 2008, because mobile was like this when it started. It could have even been 2009, but Verizon—the emperor, the empire—had to strike back against AT&T, and there was a window.

Yeah. And they went with our stuff. Look, they would have taken our stuff because they could put pressure back on the manufacturers, but we didn't have the stuff they wanted at the right time. They went Android.

And then we kept pushing because that's—I believe in staying hardcore and then learning and fixing. The problem was we were so locked into our model. It was hard to say, “Hey, we're going to learn and fix.”

Would Microsoft—I don't know where we would have gone with things on phone if I had stuck around, but I probably would have stayed at it. Maybe it would be an Android phone at this stage. Who knows? Maybe not. If you think of yourself as just a platform company, you say, “We can't do that.” If you can think of yourself as an app-and-platform company, with apps that are extensible, then you can say, “Hey, we actually have a pretty cool user experience that can also leverage some things that we do and can leverage our software skills.”

It's okay to embrace that competitor and extend. But there are so many technologies that are hard to not just popularize, but even get good at unless you have a phone these days. Just take voice. If you want to really be good at voice, you've got to get enough signal, and you get the signal off the phone. You can't say talking to my PC is sufficient.

And it's not the only one. If you want to get good at maps, there are so many things where being on phones matters. There are some things you can even make happen by being on cars. I think Tesla gets good at certain things in software because it is a different form of mobile, so they get good at different things. But we missed.

Should the company have kept after it? I don't know. That's not my call. Satya and Amy and company, they were where they were. But to your original question, for big companies deciding, well, it's not always a mistake to build off what you got, but it can be. Try to get out of inside of yourself. If you get in, do you have the ability from the top to shake the system and say, “No, we started with our old model, but it ain't going to work”?

And that's what I did with Surface. It hasn't played out, and partly I didn't have as much time with it. But there were no high-end PCs that would really compete with Mac, and I decided the only way we were going to get there—we couldn't sit there with our OEM model and have it work if we were going to have high-end PCs that appealed to users.

I wanted us to be a consumer-user company, not just an IT company. ThinkPad had—IBM, then by then Lenovo, had some higher-end computers—but you never saw them. You never saw them in schools. You never saw them in coffee shops. We needed a high-end PC, and the economics weren't going to let marketing and romancing it happen. That was not going to be an option for our OEMs. I said, “We've got to go do Surface now.”

Would we have tweaked things, done things a little bit better? Or part of that iPad? Sure. But the model was not going to work.

Okay. So we spent a lot of time talking about all these bets that sound very reasonable to make—in mobile, in search. We didn't talk about social, but in social, and all the dancing you did with Mark Zuckerberg over the years, in Yahoo, in all these things that ended up not panning out. These were trillion-dollar companies that were built not inside of Microsoft.

We talked about 1 multitrillion-dollar thing that did work with the enterprise. There's another one with Azure. Can you tell us the story of how Azure really got started?

8. Azure Starts As An Incubation

Yeah. So we are in probably 2005, 2006. AWS has a little liftoff. I think AWS comes to market around then. It's not like the cloud is some surprise to us. The Energizer, if you go all the way back to that Energizer thing from the mid '90s—it's all about the cloud. It's before it was called the cloud. It's before all the infrastructure that becomes the cloud.

It's not like we say, “Oh, woke up 1 day, oh, there's AWS.” We didn't wake up 1 day and say, “Oh, there's backends to applications, too.” We've been doing that with Windows Server and SQL Server. We've been in the cloud.

But at that point, I think we might have already had Exchange in the cloud as a standard product, which you have to remember is super important because I really want to give you my sense of what Microsoft's businesses are. We didn't have a platform.

And so I said, “We’ve got to do one. Let’s go get Cutler. Let’s just go get Cutler.” So I said, “Okay, we’ve got to get Cutler on.” Cutler and I have a great relationship. To this day, we have a great relationship. We’re personal friends. He’s still writing code at Microsoft. We’ve been to basketball games together. We’ve played golf a number of times. We’ve done golf trips together.

But Cutler’s a hard-ass at work. If he doesn’t want to do something, he’ll tell you. If he thinks you’re wrong, he’ll tell you. If he thinks somebody else in the organization is bad, he’ll tell you. He’s like a thoroughbred horse: he can run really fast, but you’ve got to get him lined up. He was a great athlete in college—2 sports. I think he played maybe 3, even, in college.

So anyway, I get Cutler, and there’s a guy working in MSR who I think is underutilized, too. This guy, Amitabh Srivastava, who you guys talk about. I thought he was underutilized doing what he was doing. So grab him, grab Cutler, and bring them both onto this project.

I think Bill—Billy—was still with the company. He was about to transition out. He was about to leave, I think.

Yeah, yeah. I think he had probably told you that he was leaving. He told me, but talk about that.

He told me, but he hadn’t left yet. So he was involved until he left. Even then, it was a different nature of involvement.

So anyway, I get Cutler and Amitabh to go do this thing. Then Cutler brings some of his—I’ll call it his gang, his favorite guys. He brings them over because he’s a magnet for talent. We get started.

We made an explicit decision. I guess you could say it’s also a function of thinking Windows first. I think you guys may have talked about this in your episode. We said we were going to build platform as a service because it’s a Windows platform.

Infrastructure as a service, if you think about it, is by nature accepting everybody’s infrastructure. It’s by nature multiplatform. You become a different kind of platform because you’re running other people’s Linux and whatever. It doesn’t leverage Microsoft’s strength of owning the Windows franchise if you’re just going to be infrastructure. It does leverage our strengths in the sense that we’ve got great low-level operating-system people. We have all the talent to go do it.

But we said, “Hey, we’re going to do platform as a service.” It was explicit. We wanted to do platform as a service. We said, “They’re doing it, and it’s all about the developers.” If it’s all about the developers, then you’ve got to have platform as a service, not just infrastructure as a service.

That assumes that the developers targeting Windows Server are still a big, strong, important, relevant developer group. They were, and they weren’t. Windows Server had a strong developer group. Unix had a strong developer group. On the front end, Windows was definitely stronger. On the back end, Unix was definitely stronger.

But on the front end, by 2006–07, the web was clearly the emerging developer platform of choice. Emerging. Absolutely emerging. Not fully emerged.

I would challenge you to say: in 2006, what amazing Windows apps were coming out that would sweep the world and get 100 million users because they were great?

Hard for me to remember. I think if you go to the field of productivity, the answer is yes. The problem is, if you left the areas of productivity and gaming—productivity and gaming, yes—if you leave productivity and gaming, I think the answer was no.

We talked about this a lot. People remember the web wasn’t good for a number of things because people couldn’t count on it. People didn’t feel like they could count on the connectivity, either the amount of bandwidth or latency, or just its very existence. We were still at that point.

So I’m not saying we were right in the way we thought about it. I’m not saying that. But I’m also saying there was still a great Windows developer ecosystem. It didn’t go from a lot in ’99 to nothing by ’05.

Totally fair.

And then on Windows Server, Unix was stronger on the back end. Of course, we were trying to make Windows strong, and we were trying to get to the cloud. Then we were learning more things about the cloud from both Exchange in the cloud and Azure in the cloud.

How do you make it easy to provision? What’s the speed of provisioning? What do you do to serve developers? There was the notion that you give them a set of free usage and then let them embrace it, because developers have 2 aspects, too.

There are developers who are not part of enterprises, and there are developers who are. The developers who are not part of enterprises need a whole different sales motion. You can call them consumer developers—not developers of consumer apps, but they are not like big corporations in terms of the way they use it. Students are an example, but there are plenty of others who are trying to do startups.

So in any event, we kind of get going. We’re learning how to do these things. We’re building capability for sure in the cloud through both products. By the time I leave, we have some momentum with Azure, but some momentum. The big momentum really is in the last 11 years since I left.

You’re bypassing and underselling here. It really struck me, as you were describing the challenges of a big company like Microsoft attacking wildly different vectors—mobile, search, hardware—that the cloud was extremely disruptive to Server and Tools. Extremely disruptive.

Yes and no. The things we understood were translatable. Now, getting the company—people get locked into a model. You had to replace servers with things that run in the cloud. That was not obvious back in 2008 or 2009.

It’s not like Amazon was an enterprise company at the time. It was mostly for startups, and that’s who was using AWS at the time. So, no, I agree. I do agree with you. We had to shake up our internal culture.

“God dang it. This is our future. We can preserve and enhance these businesses. We can take more value out of the system because other people—the customers—don’t have to set up their servers anymore. They don’t have to do all this work. Essentially, money that would have been spent on people and hardware will get spent with us. Come on, we’re going to do this.”

It was hard for me, even telling our people. There was still la résistance, as they say. That’s why I did the speech at U-Dub where we talked about the fact that we’re all in on the cloud. It was partly to remind people: get with it or get out of the way.

Making an external speech to communicate something to your internal employees—in a big company, man, I’ll tell you, it’s some of what you have to do, because people believe the newspaper more than they’ll believe an internal email.

People always talk about how the Think Different campaign Steve Jobs did was for Apple employees as much as it was—in fact, way more than it was—for the general public.

Going back to the core initial start of Azure, I find it very interesting that Microsoft had a business called Server and Tools Business, and that is not where Azure started. Azure started as an incubation by Ray Ozzie, with a completely separate team from your existing product group selling Server and Tools.

But this is sort of a classic thing that’s not mind-blowing. Windows and Windows NT were in different groups, too. Sometimes, in order to protect the baby while it grows up, you can’t put it with the thing that’s established. You could say that’s part of the issue with Windows when we tried to use Windows on things for which we probably should have started elsewhere.

I was going to ask you differently if you’d taken this approach with Windows—

We did break it out, but we constrained it with Windows. We broke Windows NT out and constrained it with Windows. It worked fine because Windows belonged.

You know how you do those incubations. In this case, I just said, “Look, it’ll probably get subsumed.” I don’t know. Partly, Ray wanted—Ray wanted some operating control over the thing, and putting it under Muglia would have made it harder for Ray. Obviously, it was less palatable.

I’m not sure Cutler would have gone to work on it if it was all Server and Tools, but it was the right thing to do, even though it was, quote, part of the future of Server. It was the future of Server and Tools, essentially.

This is pretty lost in the common narrative. If this is 2006, that’s 7 or 8 years before you left Microsoft.

Yeah, 8 years.

8 years. People think everything in tech gets popular in 10 minutes. It’s kind of like people think Acquired was founded 2 years ago. Good point, different scale. When was OpenAI actually founded?

2016, I think.

Okay, yeah. So 7 or 8 years after it really became something.

Okay. Fair to say. I give them all the credit in the world. 7 or 8 years—most things take a while. Even things that are, quote, “Oh, they just burst on the scene,” people have been sweating blood, sweat, and tears for years before these things get liftoff, as I call it in my little deck here.

So, yeah, we were starting to get to liftoff, but, yeah, 8 years. And we had more in on Exchange. Most businesses are zero trick ponies. You never create a billion-dollar business.

You might create something that goes nowhere. You might create what's essentially a feature for somebody else's business and get acquired. I'll call that zero tricks. Then you get a one-trick pony.

One-trick ponies are amazing. People should be in awe of one-trick ponies. One-trick ponies are 50- to 100-billion-dollar market-cap companies, or could be more.

Or could be more. Not many. I'd argue that Google's a one- to one-and-a-half-trick pony still. If you just look at its revenue, 80% is search ad revenue. You can call YouTube half a trick, or you can call it a second trick, but it's not clearly a second trick. They're huge, and they have a great market cap.

TSMC—you did an episode on them. They're a one-trick pony. A very successful one-trick pony. Nvidia is a one-trick pony.

Well, gaming and AI.

Okay, two-trick pony. But the first trick wasn't that big. I'll give them two tricks. You can decide whether to call it a trick or not. I'm not taking anything away from Nvidia, and I should know the company better.

But you say one-trick ponies are amazing. Everybody should be in awe of a one-trick pony.

Now, two-trick ponies—ooh la la. Those people tend to go down in business history, especially if those tricks stay alive for a long time. IBM was a one-trick pony. Microsoft: two to two and a half tricks.

All right, give us your trick accounting.

You could do it a little differently. I'm going to call the desktop business, which includes Windows and Office, and the server-enterprise business, or back office, two tricks.

Both tricks could have died if they didn't get moved to the cloud. I knew they could die, but they're two tricks: two different revenue models, two different licensing models, essentially different sales motions. Even the way Microsoft sells that stuff—I don't know about today, but when I left, they were kind of different muscles. One account manager, two different muscles.

One, you're selling applications, and one you're just selling, "Hey, this is to serve your users. You need an AD account, an Exchange account." That's exactly what you need. What Microsoft 365 and Azure are, you could call the modern translation of those two things: the Windows OEM business, Microsoft 365, and Azure.

Then you could say, "Is gaming its own trick?" I call it a half a trick, just like YouTube. It's a half a trick.

This is an update since we last talked. I feel like we had a conversation at one point where we both kind of landed in unclear territory about how profitable that business is for Microsoft. I'm going to call it a half trick, but you could say it could be a trick.

I would say Microsoft is optimistic that it'll be a full-on trick.

Okay. Oh, yeah. I hope it is.

I run into Phil Spencer at the golf course, and he's a real optimistic guy. It could be.

All right, I'll give you this: if we call Nvidia's first trick a full trick, then Xbox is a full trick.

There we go. Whatever you want to call it.

You said it's a small trick, and I think that's probably right.

That's amazing. Amazon's a two-tricker: AWS and the store. They're a two-tricker. Apple's two tricks. What's your trick accounting there?

Mac and mobile, if you want to say it's high-power consumption and low-power consumption.

Is it fair to call services a third? By my estimates, their profit dollars from services have now eclipsed iPhone hardware profit.

I consider it just part of the trick. If you go by your platform, they've just monetized it. It's kind of like us adding things to Office and redoing the EA. It's a monetization model—an additional monetization model—but it's not a new locomotive.

A locomotive is the business that can pull the caboose, and the locomotive remains the phone. The services business would go away pretty quickly if the phone volume fell apart.

I'm going to call it additional—very important, but not uncorrelated, the way that AWS and the store are. I get the sense, and I think Mac versus everything iOS is also uncorrelated.

Yep. So I get the sense you really wanted three tricks.

Absolutely.

What's the one that eats you up inside? Which one do you think you were closest to getting that you didn't get?

Not social.

Okay, forget social. It doesn't feel Microsofty. You wanted to buy—I'm going to tell you why. It's either—

Because they were still on the Paul Allen strategy: "We've got to do all the software that these things will ever need."

It was still the mindset that said—and there's an arrogance to that, and a hunger to that—there's just nothing we shouldn't do. I don't think that was a good mindset by the time I took over. And yet it was still sort of baked in with Bill, baked in with me. I think that was a mistake.

Not focus?

Social doesn't. But, you know, this is like asking me to pick between my children. I don't know. The phone, because it was a client-side device, or search, because it was a productivity tool. Microsoft was in both of those big businesses.

Yep.

The desktop, the phone, or Office—or, you know, client-side devices. We had done well with a certain model. Our minds should have been able to wrap around client-side devices, but we had to tell ourselves it didn't look the same. Its technology didn't look the same, nor did the business model.

Astonishingly, for search advertising—call it 2005—I think Google was making more money off a PC user than Microsoft was, because its business model generated more search revenue.

By 2005? I don't think so.

Not later on. I think so, but not by '05. I would suspect not. You can go check.

But isn't that astonishing? For enterprise PCs—PCs bought by businesses—it certainly would have been the case. For consumer PCs, it could well have already been the case, right?

It actually is a notable difference. Our post-sales monetization was with applications; theirs was with ads. But it was a new productivity app. We put Office on the back. By then, we would have had to put productivity elsewhere.

In the sense that we missed a major productivity area, and we're in the productivity business, and we were in the client area, and we missed a client device, those are the two.

So you feel like there's nothing else we missed? You had an opportunity for four tricks, and you got two.

Yeah. Part of the problem was we didn't see—particularly, we didn't see mobile as a different trick. We thought of it as underneath the Windows trick, if you will.

I don't know that I could come up with a three-trick pony for you.

I mean, it's possible that, at the Elon level, the Musk empire could have three tricks, right? Cars, connectivity, and finance. You can do it: finance.

I don't think there are multiple tricks. You could say asset management versus investment banking is different. Maybe, maybe. I don't know. I'm not convinced. But I hear you, possibly.

I think this makes sense because Microsoft is the most valuable company in the world with two. If anybody—

Yeah, if you look at the most valuable companies, you're not going to find three. That's a good point.

Sony is nowhere near the market cap of these companies, but it's pretty evenly diversified across its 5 segments, from gaming to consumer electronics, movies, music, and finance.

Yeah, they have a remarkable portfolio. They bought businesses in multiple areas, but I can't call Sony Pictures a trick.

Fair.

It's just not big enough.

What you can acquire to start a trick—

I mean, that part, I have no pride. There should be no pride in having a trick that starts with something small. Android's a great example. Google bought Android, and that's a trick for them.

Well, Android's not a trick. As you highlighted, Android is a piece of the search trick. It's lead generation.

Yeah, exactly. Lead generation for search. That's right.

Okay. So, we've been dwelling here in the products and reflecting back on big wins and misses. During your tenure as CEO, can you reflect back on your nonproduct wins and mistakes?

9. Ballmer Runs The Company Through Crisis

Look, my biggest hit from my time running sales to president to CEO was establishing us with IT departments and IT professionals—you can call that the enterprise, if you will—and putting in the framework from a sales and marketing perspective, the staff. It's a capability we had to develop. Nobody developed that software model but us. We invented essentially how you do that.

Oracle had done some invention, but we came on and did our own invention. We took it to the cloud. We were able to successfully navigate that. I mean, look, from a sales perspective, there's a product part to that, which you highlight, but that's a big deal, and I feel very, very proud about that.

From a financial standpoint, everybody likes to say we about tripled revenue and tripled profit. The truth is, we dramatically increased profit by more than triple because people forget there was a major change that came along early in my tenure, and that's the move to having to expense stock options. So, if you had restated our books to the time I actually took over, stock option expense would have reduced profits notably.

Stock options were unaccounted for. If you look at what starting profitability would have looked like if stock options had been expensed, it would have been lower, and the multiple over my tenure would have been much more than 3.

Okay. So, 3-plus. I think you might say 3 in revenue and probably closer to 4 or 5, maybe even, on profit, about the same time the dot-com bubble bursts. So, you have 2 problems. Number 1, now we're showing our books with all this expense for stock options.

Okay, but people don't value those things that we have to expense, and the stock is flat, so they value them even less. This is a really insidious problem. You've got to get rid of stock options.

We transitioned then from stock options to stock awards, which, if you notice, I think we were the first to make that as a major transition, but everybody's made the same transition. With the exception of a few senior executives, options are not the primary form of compensation. It's a little different in startups, but when you look at larger companies, everybody—even startups—is now doing RSUs. We had to start that. I didn't realize that Microsoft started that—you can check—but I know we moved before most of the tech companies.

It's a tough thing to have to inherit right at the beginning of your tenure, coming off of an already all-time-high multiple of the stock price. The dot-com bubble bursting meant our stock price burst, too. But I think, to your point, what you're saying is this became an employee-motivation and cultural issue.

It's not just that we had 2 problems before the dot-com bubble burst. You had everybody saying, “Oh, maybe we should go to a dot-com company because we're going to make a lot more money.” Then the bubble bursts, and everybody says, you know, sort of, “You guys—the movie Oklahoma!” But there's a song, “Poor Jud Is Daid.” Poor Jud is dead. Poor Jud is dead. Candlelight is dead.

Absolutely. Candlelight is dead.

And that was kind of the way people felt about stock compensation. And not just at our place—people were down because everybody thought they had a ton, and then they thought they had less.

So, yeah, it was a real employee morale issue in the early 2000s. We had to really sell this stuff in. That's a big thing I had to work on. Obviously, the antitrust issues.

When you took over as CEO, what we said in our episodes was that your number-one priority was to end this.

It was right up there. I think when I took over, I'm not even sure we saw a path to resolution, but having it as an overhang—

I'll give you a story. It was after I took over as CEO. We had an executive retreat. We did it down in Bend, Oregon. I can't remember the name of the lodge—Sunriver, I think. We all flew down there. We rented a plane to fly everybody down there. I don't know how many people there were by then. It was probably 80 or 90, something like that.

The first session was supposed to be a report from the field: What are people seeing out there? What's the environment? This guy, Orlando Ayala, was running sales at the time. He gets up and—you know, this is probably 2002-ish, 2001–2002. We're still in the throes of the thing—and says, “My name is Orlando Ayala. I am a proud Colombian. I am not a proud Microsoftie. Today, our integrity is under assault. My personal integrity feels like it's under assault.”

Now, he didn't blame us for having behaved badly, but he highlighted the thing that was on everybody's mind: It wasn't just a business issue that needed to be taken care of. It was a cultural issue. It was bothering people, particularly senior people, very personally.

I had this whole agenda and had to blow the thing up and reorient to address that elephant in the room. It was not where I was going with this thing. I had to completely remap it, change the breakout sessions, and focus in on this issue.

Bill was not happy with the whole thing. Bill bore the weight of the antitrust thing very hard because, for him, I think it also felt like a personal attack, of course. Everybody took it personally. Bill took it even more personally because he was the face of vilification, if you will, for this.

But it's a reminder that it was a cultural issue to take care of, not just a market issue. People focus in on, “Oh, were you moving slowly?” Yeah, there was some of that, too. People said, “Oh, I wonder if we can do this.” That was an issue. The cultural issue, I think, was even bigger.

And then there was the order to break us up. I forget what year that was. You were going to run 1 company, and Bill was going to run the other company.

Yeah. We never really got to the point of really planning that through.

But that's what the federal government ordered, right?

Yeah. No, they ordered it split. They didn't say who had to run which. I think it was just that you couldn't be at the same company. I would run operating systems, and Bill would take applications. It just gives you a sense of what each of us was associated with in the mind of the company.

So that's your starting place as you're taking over as CEO: the dot-com bubble's bursting, antitrust is dominating the company's culture and the external narrative, and you have this big accounting headache that you now have to deal with that affects the way your profitability is shown. But then there's a decade after that where you triple the business, but the stock price is flat. Why didn't Wall Street get it?

I'm going to give you 3 reasons. Reason number 1, and it's material: Bill and I always—we were always trying to tell people, “Don't get our stock price too high. Don't have too big expectations for us.” We never wanted people to feel like they got cheated buying our stock. Partly, probably, we're trying to lower the expectations on ourselves. I never thought of it that way.

I never thought of it that way.

I don't think Bill did, but essentially that was part of it. So, we do this financial analyst meeting every July, and we'd always warn people, “Don't get too excited.” That's 1.

As part of that whole theme, Bill never went to a quarterly analyst call, and I never went to a quarterly analyst call. If you really think about it, part of morale is the stock price.

It is. It took me a while to realize that, but I never broke my pattern. It's sort of like going to the newspaper every day. You don't sell stock every day, so you really should only care what the stock price is when you sell stock. But people go every day, and it's kind of like, “Oh, did my team win last night?” It's like going to the sports section and saying, “How did the Clips do last night?”

Talking more regularly to investors and talking with not a pie-in-the-sky but a realistic view of guidance—we gave no guidance. I had to fight people. They wanted to give guidance, and I didn't want to give guidance.

Why?

Just deliver the results you get. There was a bit of a Buffett-style thing going on because Bill and Warren were very good friends, and Warren didn't go to quarterly calls, I don't think. But, you know, he's Warren. I don't even know if they do quarterly calls. I don't think they do.

If they do, it's the annual meeting, obviously. So, that's—let's call that a first reason.

A second reason is, yes, I did take over when the stock was ridiculously highly priced. But that normalized within a year or 2. The bubble burst. It normalized some, but it created another narrative about things.

Well, no, I'll give you 4 reasons, then. Next, I was hardcore about telling people, “I'm going to spend to do the things we need to do to succeed.” That's not what Wall Street likes to hear.

No.

But I was viewed as a spender, and I was much louder on this than Satya is on anything financial because it's kind of how I'm programmed. He's programmed a little differently. And Amy is more balanced.

I mean, she'll talk about balance, and I would say we're going to win with Surface. I mean, whatever it is.

If I could paraphrase my view of it, you were willing to say, “We're going to spend whatever it takes.” And Amy goes and says, “I'm gonna account for every dollar of spend real tight and make sure that every dollar demands a return.”

Yeah. And so I had no credibility in terms of what some investors wanted to hear. My actions were consistent with that. It's not like they were inconsistent.

Lastly, people did worry about the future of a couple of our franchises, most notably Windows. So you get all these things: a narrative transition from high price, some issue about franchises, and me being a big spender. No wonder the stock stayed flat, and by the end of my tenure, it was even bothering me.

When did it start to boil toward the end?

I mean, at some point, I just got too tired. But by then, it was also probably hard for me to reset that dialogue, for me to go to investors and say, “I'm a changed man. I'm not going to spend anymore.” Nobody was going to believe that. They just wouldn't have believed it, right?

You can't come in and say, “Well, I've been a certain way for about 35 years—or 30 years—but, hey, I'm a new man. I'm reformed.” It doesn't work that way. If you're a spender, you're a spender. I worship at the altar of capital allocation now, but if you're a spender, you're a spender. If you're not good with investors, they're not going to buy in overnight that you've changed.

You know, there's a certain—and I didn't intend it that way—but there's a certain disrespect by not going to quarterly calls. With hindsight, people aren't going to say, “Oh, he's showing up. He's a changed man. He used to tell us the stock price was too high or worry about it. Now he's going to tell us, ‘No, the stock should be okay. It should be higher.’” No, there was no way to reset the investor view of me.

You need a full rebrand, a full clean slate.

Well, you probably need a full new CEO. When I wrote my letter of goodbye to the board, I did say, “Hey, look, this is a unique opportunity. There are a lot of things in our brand and in our image that would only be able to be reset by a new CEO, by having a new CEO, because people don't walk in and say, ‘Oh, yeah, these guys are changing.’” So it's hard to change the narrative without the change.

Now, I'm not saying that means the CEO should go every time there's a bad narrative. That's not really my point. But it just gets harder, particularly since, look, I might have only been CEO since 2000, but it's not like I wasn't there since 1980. I was there since 1980, and essentially, I'd been the second voice of the company for 20 years. Then, for 14 years, I was the first voice, theoretically, although that had some complexity too.

I kind of get the sense that by the end, it wasn't fun for you anymore either.

No, that's not true. No, no. Look, the toughest time was probably shipping Vista. That was probably the toughest time, and the early 2000s, when I took over. On my little sheet here, I highlight that 1998 to 2004 were kind of tough years, plus Xbox, because of the antitrust case.

That's where I moved back to be president of the company, then CEO, and Bill and I went through a year where we didn't speak, really.

I think it was basically from sometime around March or April 2000 to 2001. I mean, literally, we weren't speaking. I didn't know what it meant to be his boss, and he didn't know what it meant to work for me.

When he asked me to be CEO, I said to him, “Look, and I know you're struggling with the DOJ and all this. Do you really want me to be CEO, or do you just want me to be a figurehead?” And he said, “No, I want you to be a real CEO.”

That meant something to me. I probably would have said yes even if he'd said, “Be a figurehead.” But he said what he wanted, and probably he was saying to himself, “Hey, I've got to have a transition path.” So I said, “Okay, I'll do that.”

Well, he didn't know how to show me a different kind of respect, and I didn't know how to show him a different kind of respect. There were things where I just disagreed with him, and now I expected it to go the other way.

I was always happy being a number-two guy. It was fine: “Salute. I don't like the decision.” I either saluted, or I'd body-punch and then salute, or body-punch and he'd agree with me. Body-punch means it's a slower process. And then, you know, we didn't know how to do that. We just didn't know how to do that.

After a year, we started talking again. Basically, our wives were the ones who pushed us back together. We had a very awkward dinner at a health club down the street here. But we got back together, and we never really got the right mojo.

I mean, Bill was chief software architect, and I was very deferential then to product direction from Bill. He was working on Longhorn at this point because it was post-XP, which was a mistake. Longhorn was a big mistake. I have to take accountability; I was CEO. Bill's got to take a lot of accountability. It was the mistake of mistakes.

Between the company, Bill, and me, we disagreed about whether we should do hardware. That was a big one. Surface was a big disagreement. Phone, big disagreement. HoloLens, big disagreement.

What about Azure? Were you aligned?

Bill was fine with Azure. The cloud, Bill and I had agreed on in the '90s, right? I mean, Azure—Azure, I think Azure could have been Bill's idea, not mine. Yeah, pretty sure it was Bill's idea, not mine. I executed, but Azure was Bill's idea, not mine.

But we never hit it. There were places where there should have been more contention, maybe even during the late '90s. I don't know, but there were certainly places where there should have been more contention.

And my gut was, you know, these are the smart technical guys, Bill and some others, and I'm trusting Vista. I'm beginning to have a pit in my stomach, but we didn't have the right contention. I mean, was it—and this is not directed at Bill personally; it's directed at all of us—we kind of had an emperor that had no clothes.

Yeah, Longhorn was the emperor that had no clothes. Partly, it was the centrality of Windows and the notion that Windows would stay central, therefore people would all want this new stuff. Partly, there was too much change all at one time.

We didn't do a new operating system, but we were kind of doing a new operating system. We probably would have been better. It may not have sold at all, but we probably would have done better just doing Windows.

Yeah, yeah.

No, forget what we called it. Just starting from scratch. Maybe keeping parts of the kernel, but otherwise starting from scratch and throwing out all that code—all the cruft.

Now, I don't think we would have popularized it. If we'd looked at it that way, we probably wouldn't have built it. But by then, we were a little cocky about Windows, and it was our thing. So I don't think we had the right grind in our system there in the early 2000s.

Just between Bill and me, did we make some good decisions? Yeah, we did make a good decision to do Xbox. Were we doing too many things? Yeah, we were doing too many things. And I would say there was probably a voraciousness misplaced by Bill and me.

I had to deal with some of the pragmatics of hiring people and stuff, so I probably didn't push back on it, but I probably felt the pain a little bit more in terms of trying to hire people. So that's kind of 2000 to 2004.

And then by 2004, Bill was already sort of talking to me about wanting to be able to go. In 2006, we announced that he was going to go in 2008. I also think we screwed that up. You can't have a long goodbye. Long goodbyes are not helpful.

Yeah. Yours was short.

Yeah, it was goodbye. I stayed on the board for one more board meeting after I left. That was it. But a long goodbye—then nobody knows their role.

I think I did some of my very best work after Bill left. If you ask me when I think I did my best work, it was when I started running sales and sort of evolving this enterprise business, when I ran system software, and then the last 6 years I was there.

That's cloud, that's Surface, that's some of the improvements in Windows. I feel really good about my last 6 years there. Bing—that's when we hired—I think that's when we hired Qi Lu.

Qi Lu?

Yeah, Qi Lu. Qi Lu was one of the most pivotal things at Microsoft.

Why? I knew he was important, but tell us the story.

Pivotal in a way you may not even know. First of all, a brilliant guy, a great guy. He's talking about leaving Yahoo. He's at Yahoo at the time, and I think he went to graduate school with Harry Shum, who had been in Microsoft Research. Harry was now working on search, and he was working for Satya, who was running Bing.

Oh, Satya—the guy who was running Bing.

He was running Bing. And Harry says, “He's a genius. We've got to hire Qi. We don't know if Qi really wants to work, but we've got to pick Qi's brain. We just have to learn from Qi.”

So Satya, Harry, and I flew down to California, and we met with Qi and talked to Qi. Qi's brilliant. We're learning all this stuff about Qi, and Qi leaves the room.

“God, there's a lot.” I don't know who throws the idea out at first. Maybe Satya: “We should hire Qi, and I should work for him.”

Whoa. Harry was all in. Harry worked for Satya, who worked for Qi. Now we flipped it around.

You flipped the whole reporting structure to hire Qi in the room.

After Qi walked out, we talked for about 15 minutes, and then Harry called Qi and said, “Do you mind coming back?”

Wow. Wow.

I forget where Qi was thinking he’d take his next job. He had a next job in mind. Maybe it was with Baidu; I can’t remember. Someplace.

So then what did he do at Microsoft that made him so impactful?

It’s the story I just told you. It’s what it told me about Satya. I loved Satya. We were giving him more and more responsibilities anyway. But it told me this guy will do the right thing for the company. He’ll prioritize that. He doesn’t have an ego that gets in the way.

Qi did great work. Qi knew about search. He was an old pro at it, and it started cash-flowing billions of dollars eventually.

Eventually.

I mean, Qi is an engineer by training. He’s a PhD in computer science, and he had a lot to bring. Satya has been great at managing product development, that’s for sure. But Qi is the one who’s digging into the bits and bytes.

Sure, but the meeting is the thing that was important. Satya was important, sure, but what Satya and Harry did that day—they just found a guy and said, “We’ll hire him. Please, Steve, go hire him as our boss.” You don’t hear that very often.

No.

What year was that?

Let’s see. What year would Qi have come? It was probably after Yahoo.

Yeah, 2008, 2009.

Six years before 2014.

Yeah. Satya became CEO in 2014, or 5 years later.

And that let me then be able to say, “Now I can give Satya more responsibility doing something else.”

Why did you move him to Server and Tools?

I thought it would be great. We had Qi, so we could probably move him. I thought it would be important to give him other experiences to try to get him to be able to be CEO, because he was on a list of 3 or 4 internal candidates at that time. We had been talking about it because we did an annual succession-planning exercise.

Succession planning has 2 candidates: What happens if you get hit by a bus, and what happens if you serve your term, whatever term that feels like? They’re different people, right? If Satya gets hit by a bus, if Satya serves another 5 years, it’s probably a different person. I think that’s true in most companies. You’ve got to think about it differently.

Anyway, I said, “Hey, we’ll get him another experience.” He hadn’t worked in apps, and he hadn’t worked in Server and Tools. It was kind of a good time to switch things around.

Bob Muglia was obviously being super successful because Bob was running Server and Tools at the time. I love Bob. Bob’s one of my favorite guys I’ve ever worked with. He went on to be CEO of Snowflake.

Snowflake. Yeah, absolutely.

Worked out for everybody. He’s done fantastically well.

But we moved Satya into that job. He was on a great path, and Qi’s hire made search stronger and stronger. It showed just how right Satya was.

We talk about this in basketball. Is it all about team first or not?

All about team first, which is essential. We were able to give him the additional experiences, which were super helpful in terms of him then taking over as CEO.

That was turbocharging.

We want to talk a little bit about your post-Microsoft term, but let’s leave Microsoft with a final question: Why did you resign?

10. Ballmer Leaves Microsoft

A couple things. 2 or 3 things. Number 1, the phone was very much on my brain when you said, “Are you having fun?” That was the thing that was eating at me the most—the phone.

I decided we needed to flip the model around. Your episode is pretty good about all that happened, so I’m not going to go through all that. But I knew we had to do hardware. I knew it. There was just no question.

We weren’t going to be able to play the search game, the Android/Arch game, because we just didn’t have the power of monetization that they did. And Apple’s Apple, but there are going to be 2 phones. It’s not like there’d only be 1 phone that was popular in the world.

This is something you guys didn’t put in the episode. I’d been trying to buy a hardware company. I’d talked about buying a phone company for years, a number of years before the Nokia deal. I forget what year it was. I flew to Taiwan, and we were looking at buying HTC. They were the biggest Windows Phone OEM at the time. Nokia wasn’t signed up.

I finally just decided—Terry Myerson and I had 3 or 4 trips to Taiwan to talk to Peter and look at the organization—and I decided it would be too tough to buy a Taiwanese company. I would worry too much about the integration. I liked Peter Chou, who ran HTC. I don’t know if that name means anything to you guys.

Yeah, of course.

I’d been looking at that thing for 2 or 3 years, maybe, before. Bill and I had, and continue to have, all the tension we had about anything that had hardware in it. It’s not like our relationship was calm and clear. It had always been bumpy.

Even back to the beginning, I almost quit after 4 weeks—5 weeks, maybe—because we were in a fight. It’s not like it had ever been linear. That would have been a very poor economic decision. It had never been linear. It had helped build Microsoft, but that didn’t mean it had always been easy for him or me.

The hardware thing was exacerbating our relationship. I thought we really needed to do a phone. Then the board said, “No, we don’t want to do a phone.”

I was very transparent with everybody. We brought the management team in. I don’t know if more wanted to buy or didn’t want to buy, but I let everybody speak. It’s a big decision to be in the phone hardware business.

The process from when we did the presentation to when the board said no, I didn’t find very respectful. The board didn’t ask me to leave; I just didn’t find the process very respectful. I probably won’t go into the detail of that.

A lot of it has to do, again, with my relationship with Bill because we’re—and look, I knew Bill didn’t love the idea, and I was willing to accept whatever the board decided. No question about that. But the process wasn’t very good. I was not happy with the process, and they wanted me to stay.

But I decided 2 things. If we’re not going to buy phones, that’s kind of my best shot for a consumer future for the company right now. That’s my best shot. I tried the Yahoo shot and the phone shot. Those were my 2 things. Remember, mobile and search.

So I said, “Look, this might be the right time. We can’t make my play here.” Not out of pique. I had thought about this in advance and said, “Look, if it doesn’t work, it doesn’t work. If the board doesn’t want to do it, fine.” And so I said, “This is a good time.”

It’s also a good time because the cloud’s just coming on. I’m saying to myself, “Look, we’re going to have to build new capabilities. Even the way we’re moving from a 100% gross-margin business to something below 100% gross margin, we have whole new capabilities we need to build up around that.”

I even think of it through the lens of the accounting system. We have these revenue and cost reports. They have to change in the world of the cloud because you really have to get tight on gross margin, not on revenue.

I don't really pay much attention to Microsoft's revenue. I pay attention to gross-margin growth these days. In those days, when I said the move to the cloud, I used to say this to analysts: “You should expect us—you want us to have lower gross margins going forward, but we'll make it up in volume,” right? I mean, that is the whole proposition.

Lower gross margins—it’s like Walmart’s an okay company, even though its net margin is whatever, 1.5%, 2%. You just have to make it up in volume. So I knew it was a good time to let the new person sort of build from what we had to the next generation of all the machinery that would have to happen to make cloud happen.

For one, I never lost my desire to be an end-user company, slash consumer. It killed me. And it’s sort of—you don’t just say, “I want to be a consumer company.” No, you’ve got to find the locomotive, not just a bunch of cabooses.

At the end of the day, Zune was a caboose. A lot of the things we invested in were cabooses. We had to find the locomotive. There were only 2 possible locomotives that made any sense. I didn’t have a play that I thought was going to break through anytime soon in search.

Mobile was going to be really hard, but I knew in my heart of hearts that without physical hardware, we weren’t going to break through there either, because of search. The board said no. I said, “Okay.” Bill and I are—it’s not really the board being disrespectful. Maybe it is, but it’s mostly me and Bill. We’re grinding, grinding, and that’s never fun when we grind.

And I say, “Okay, we’re grinding. I know it’s frustrating for him; it’s frustrating for me. We’re grinding. Here goes my idea. Here goes my idea.” And, oh, by the way, this is a great juncture point. So I said, “Okay, I’ll pass,” and then the board changed its mind.

So why did they end up buying Nokia, then, after your decision was final? You were out.

Oh, I don’t know. Maybe you don’t know. I don’t really know. I mean, I’m not sure they really understood what I had told them about how we had a deep partnership with Nokia. I’m not really sure the guys really understood. I had done a good job explaining how close the partnership was.

So there was really no going back to Nokia to see if we could have a bigger partnership. The problem with the partnership with Nokia is they didn’t have the money to invest in marketing. We did. They did not have the ability to go deep-pocketed. We did, but if we didn’t have the monetization capability back through the phone, we weren’t going to be able to make it work as a partnership, because we had to put in the cash and therefore we had to get the return, and it wasn’t going to work.

And it reached a point where you had to buy the company or just cut bait totally on the whole.

Yeah, just because the math wouldn’t work. What we had to do to be successful was beyond their financial capacity.

But if we were going to do what it took to be successful, we couldn’t do it on, like, $4, right? You needed the margin dollars from the hardware, too.

Exactly. Exactly.

So you left. You did a pretty incredible thing—or really, you didn’t do an incredible thing. You held everything. You’re still the largest individual shareholder in Microsoft. I think you might be, other than index funds, the largest institutional investor, too. Basically, besides Vanguard, you’re in that category.

On the one hand, I imagine that was very simple, and you’ve given reasons in other interviews in the past: you’re a loyal guy, et cetera. Just talk us through the emotions and thinking about that, because I imagine that was not so simple.

No, not. I leave, and then what does it mean to emotionally detach? Because if you’re not there, you have to emotionally detach. You can’t say—because you don’t—you can’t control anything anymore. So it’s hard. You don’t want to stay quite that emotionally attached, because it’s like, “Oh, I’ve got to get back in and fix everything.”

But I said, “I’m going to be the best investor. We’re going to know everything about this company. We’re going to go to—I’m going to read everything just like I used to. We’re going to go to conferences just like we used to.”

I went to 1 shareholder meeting and I was kind of a dick, in my opinion. I mean, literally 1 shareholder meeting, and I was too emotionally attached. So it took me about a year to say, “I just have to emotionally detach.” It took some work, but I kind of was able to get there. But I’m still loyal. I didn’t want to sell.

Then we get our philanthropy started, and I do need to do something because we do need some of the asset value to give away. So I went through a bit where we gave some away—that is, we put it into our donor-advised fund. I also sold a little bit at the time, and I was thinking this was around 2015-ish. Might have been even 2016, something like that.

Connie had been giving away money, but the dollar value was ramping up. Then I said, “Maybe I should just sell it all. Full emotional detachment.”

Wow.

Let's do full emotional detachment, because look, it was my baby. It's my baby. I mean, I'm not a founder, but I think of myself as a founder. I was there so early, and I hired basically everybody. Everybody who was a senior leader, I'd recruited.

It's not true anymore. Now things have changed. There's probably only 10% of the people who are there now who were there when I was there, or something, at the senior levels. I mean, I can go through the math on why that's true, but that would have been a very understandable decision. It's just nothing to do with money.

And you're ready to hit the button. You're ready to hit the—

My only thought process was emotional detachment. I was wrestling. I was wrestling. And then a woman who works here—an ex-Microsoftie who works here in finance, who's the woman who sort of really charts what's going on financially at Microsoft—she and her boss, who's another ex-Microsoftie who used to work with me most closely on the financial stuff, said, “You can't sell. You can't sell. This is going to be worth a lot more. You can't sell. You can't.”

So she effectively made a Microsoft stock pick. She was recommending it. She has loyalty, too. It's not like we have a bunch of Microsofties here, and it's not like they lack loyalty either, but it was a little bit loyalty and a lot a stock pick.

I said, “Look, my loyalty trumps my emotional attachment. I can get through my emotional attachment, but my loyalty.” And look, I think of the thing as a 2-headed hydra. I thought about this the whole way. Things could go to nothing, or things could explode. That's partly why we tamped down the stock, because we always saw the possibility for either of 2 radically different outcomes.

And then finally I say, “Look, I'm not going to sweat whether we're going to get the downside or the upside. I'm just going to be loyal, and I'm going to be enough emotionally detached for this to be okay.”

Because for you, it kind of doesn't matter. There's not a downside that could be so bad—money off the table—that your family's financial security would be threatened. You could still run one of the best philanthropies of all time.

I mean, Connie would have been okay with it. She finds it difficult to give away as much money as we have, so she wouldn't have minded a smaller problem to start with. She would have been okay.

I've been charting it over the last 3 years. You guys are giving away almost in the neighborhood of $1B a year, cash out the door. But your net worth is ballooning every year way faster than you can give money away because of the Microsoft holding.

Yeah. And one other thing about the Microsoft holding that's important is the size of the dividend check.

Ah, the dividend.

Between Microsoft and the other stuff I own, the dividend checks are pretty close to what we give away.

So you can look at the appreciation, but we're just above the dividend tax. You're just trying to shovel the money that's coming in the door out the door. So you can fund the whole philanthropy without selling additional shares.

Well, there's 2 things that are going on. One, the dividend checks are pretty good. And number 2, I do have stuff that's not in Microsoft.

So you hold, I think, mostly index funds outside of the Clippers. Is that right?

Yeah—Clippers, arena, index funds. I have 1 business I invested in with a guy who I went to college with who worked at Microsoft. It's called Stagwell Media. It's a marketing-services company. You could call it a modern-day ad agency, but it's not really an ad agency. It's run by a guy named Mark Penn.

So I do have some money that's not in index funds, but mostly I'm in index funds.

Which, I mean, anybody else in the same couple of top pages of the list that you're on, you must be the only one who operates like this. Everybody else has huge family offices, lots of investments, private-equity funds.

Yeah. But if you look at the guys, I mean, look, I would say you probably would find that Zuckerberg is pretty concentrated. I don't know this, but I'm going to guess you would find—I don't know about Ellison, but obviously some of the guys who own more privately held businesses are pretty concentrated, too.

To be concentrated.

Yeah, to be concentrated. The Google guys, I imagine, are concentrated, but I don't know that. I can't speak for anybody else. Obviously Bloomberg is concentrated.

Right. Right. So, well, I think in practice it all works out the same way: there's 1 thing that is everything.

And look, if you sell it, you’re just going to pay capital gains taxes. So, if you’re really just being a financial monster about it, you’ve got to decide: will Microsoft underperform the index by enough to offset capital gains taxes? I don’t need the money. I have plenty to live on without selling anything. That’s number one, financially.

Where’s that money going to go? Some will go to my kids, but most of it is going to go to the government or to philanthropy. So why would I sell, so we have less to give to philanthropy someday? Unless I really think Microsoft is going to underperform the market by essentially the capital gains rate.

I feel like I’m watching a live Q&A video right here.

Yeah, I got this question once. I’m a member of a country club in L.A., and one of the things country clubs do sometimes is Q&As with members to entertain. I did a Q&A with a friend of mine at the club who had been president of the club, actually, and also kind of knows Charlie Munger pretty well. Charlie Munger’s there as well, and Charlie Munger comes up to him beforehand—and to me; I know Charlie through Bill and Warren—and says, “If you call on me, I have a question,” as only Charlie can.

So you did a Charlie episode.

So we do the Q—we do our panel thing, the two of us—and then Q&A, and Charlie gets up to the mic. He’s not moving super well, but he gets up to the mic, and, “Oh, Charlie, we can call on you.” Charlie says, “Steve, you know, I’m wondering why you held on to your Microsoft stock when your partners over there didn’t. I know you’re not that smart.”

I said, “No, Charlie, but I’m loyal.” I don’t know why Paul and Bill didn’t hang on. I don’t know; you’d have to ask them. But for me, it’s sort of a from-the-heart kind of thing.

And, you know, I think it’ll be fine. I don’t think I’ll screw anything up financially. I mean, what’s the worst thing that happens? Microsoft goes to zero—probably not. But even if Microsoft goes to zero, me and my family, we can live, we can give away money. It’s not going to go to zero, and I’m okay either way it goes. I’m fine.

And are the Clippers and the Intuit Dome fully paid off at this point?

The Clippers are fully paid off. I paid them off the day I bought them. That’s not true. I didn’t want to sell stocks at the time, so I borrowed some money, which is long paid off.

On Intuit Dome, we borrowed some money against Intuit Dome, so I don’t owe any money on it. Oh, that’s not true. I owe some. I have some margin debt that I used, but again, it’s just a timing thing. I didn’t want to sell stock, so I took some margin debt, which, as dividends come in, I’m reducing. But the building itself has debt on it.

Why? Because if something were to happen to me and Connie, my wife, had to sell the building, it obviously has a lower value. The buyer would have to come up with less cash because it has debt on it. So call it worth X billion, right? You just know it’s got Y billion in debt on it. You’re only selling it for X minus Y; you’re not selling it for X. Meaning, the universe of buyers is bigger because it has debt on it.

And, oh, by the way, I happened to get the debt at a very good time, at a very good rate. So it’s sort of a double value to a potential future buyer. So that’s the reason we put debt on the building. The margin debt was just a timing issue, if you will.

11. The Clippers Become A Product

I feel like I’ve done you—or we’ve done you—a great disservice by going into the Clippers and Intuit Dome through the financial lens.

Yeah.

Can I ask you now that I know you don’t own them for financial purposes?

Well, hey, and I will also tell you, unlike Microsoft, it cannot go to zero.

Yeah, yeah. Like, the asset value?

Not a chance. It is far more secure than Microsoft.

Why? They’re not making more of them?

They’re not making more of them. And as long as anybody in the world is getting richer, the buyer pool will only go up. People don’t buy them for their earnings. I wish we had more earnings, but at the end of the day, people are buying them because it’s almost more like a piece of art.

I mean, not everybody. Some people don’t like negative cash flow, blah, blah, blah. But at the end of the day, the Clippers—we have the best market in the world. I mean, you don’t want to own a basketball team anywhere other than maybe Miami. The place players want to play is L.A.

And if you look at buyers, if you’re a buyer, where do you want to go? If you don’t live in L.A., where do you want to go? Well, you want to go to L.A. or you want to go to Miami. You don’t want to go to New York in the wintertime. If you’re a foreign buyer, potentially you want to go to L.A.

We should get on to something other than asset value. I’m not selling the thing. My estate may sell it. I don’t know what Connie and the kids will want, but at the end of the day, that one does not have a lot of volatility in it. It’s a nice retirement fund.

What’s been the most surprising thing in your Clippers journey?

I’ll give you 2 parts of the answer. First is how I relate to that business versus the businesses I’ve known. Number 1, there are more similarities than I ever thought.

We do version upgrades just like you do. What’s a version upgrade? You do major, major version upgrades over the summer. That’s the draft and free agency and trades. And you do a minor version upgrade at the time of the trade deadline.

Yeah, yeah. It’s very simple. You’ve got a 6-month ship cycle. It’s your service pack. You have a major release, SP1 and SP2.

And, oh, by the way, you know how people like agile development now? Guess what? That’s called changing the game plan. The coaches are always modifying in that sense. So it’s a little bit similar.

I never thought about that.

The business is just like Microsoft. We sell both advertising, and that’s called sponsorship, and we sell tickets. That’s like software licenses. And we have an OEM business; that’s called broadcast revenue. It’s remarkably similar, 100%, I mean, just in terms of business modeling.

We do have a union, which is very different. That means complexity through the collective bargaining agreement. It also covers things like what’s the maximum salary, what trades can you make, all that—very different.

You actually are business partners with your competitors. That’s different. You actually get together and talk to them. I never did that when I was at Microsoft. But you get together and you talk to them while you’re trying to compete.

If you have somebody who wants to advance through their career, oftentimes the best way for them to advance is they have to go to another team. I mean, we have a president of basketball operations. It’s not an open job, and I don’t plan for it to be an open job. I don’t want to lose anybody. But a lot of the career moves people make would be to other organizations.

We don’t like that, but we want to have the talent everybody loves, right? At Microsoft, your domain is always growing, and so there’s always the domain growing or the number of people. You can move people: “Oh, you’re an engineer. You’ve worked on X. We’ll move you to work on a different product,” for example.

It’s different, the way you think about people, primarily because of the union, but also there are only 30 head-coaching jobs. There just are. So if somebody wants to be a head coach and they’re not our head coach, they have to find a job someplace else.

Again, not what we want. But the reality is we don’t want people held back in their career. It’s not like Microsoft, where I felt like I could always find a job that somebody should want.

I’ll give you another one to think about. Business likes to say, “Oh, we’re accountable. We’re agile. We’re this. We’re that.” Sports is so much more accountable than business. It’s like a joke. I’m being a bit extreme for fun, but every 24 seconds you get a report card: basketball’s shot clock.

Every 48 minutes, you can’t say, “I’m going to make it up next quarter. We missed, but I’ve got it next quarter.” No. You lost that game. That game is on your loss column for the rest of the season. You cannot dig yourself out of that 1-game loss hole. You can’t. It’s gone.

And you can probably also be reasonably confident about each individual’s contribution to that win or loss. Your customers know everything you know. It’s not like you could say, “Well, back in the lab, you wait until you see what we’ve got in the lab.”

No, every statistic we have, our customers have. You want to know how many miles James Harden ran last game? It comes out of the statistical systems; you can find that out. If you want to know how many pick-and-rolls we ran of a certain type, how they were guarded, and how we scored against them, don’t worry—you can read about it.

You want to look and see what the dynamics look like on the sidelines? You can just sit there and watch our players and say, “Oh, I don’t know everything. I don’t know what they’re saying, but I can see their body language. Oh, so-and-so seemed fairly charged up. Oh, that’s great. So-and-so cheers for their teammates. So-and-so seemed down.”

There’s almost nothing. I mean, we get to watch practice; our fans don’t. But the level of accountability is so high. The speed is high. Think of teamwork, man. It’s all on display.

Not only is it on display, but you absolutely know you need teamwork. One star cannot bail you out. You may have 1 star, but then the pieces have to fit around the star. It’s just the way it is.

You know how, in businesses, everybody wants to talk about teamwork? In a lot of places, that would mean, “Hey, Ben, I don’t know, we could work better on this.” And then Ben can say, “Your team’s doing things wrong.” Then we can get back together and talk a little more, and a month later we can talk about it some more.

Probably you've seen this in some organizations, and then at some point we'll talk about it as if we're a great collaboration between our two teams. You know what has to happen in our business every minute? You have to actually say it: “Pass the ball,” or, “Hey, this isn't working. You've got to do X.” You've got to give real-time feedback. You can't lollygag: “Well, you know, let's rub each other's belly.” No. If you want that team to be better, you have to hold one another accountable.

Not just the coach. On the best teams, the players hold each other accountable. And it's not just the best player holding everybody else accountable; the guys who are not stars have to be able to hold everybody accountable. Everybody's got to hold everybody accountable, which really means giving the feedback. At Microsoft, we got rid of the value called “teamwork.” I didn't want that one. I said, “Open and respectful, and dedicated to making others better.”

Because “teamwork” could sound like, “Treat everybody nice. Nice, nice.” Open, yes, you've got to say what's on your mind. Respectful, yes. But number 1, dedicated to making each other better, which I think is what the purpose of teamwork is, as opposed to the word “teamwork.”

Oh, that's interesting. Teamwork is an implementation detail, but that's not actually the goal. We don't seek to have an organization full of teamwork. It's teamwork because we want some output.

Exactly. And I think back to the old HP team: “I'm okay, you're okay. Let's all be nice to each other.” And a little bit of that has come back into the general narrative of culture today. But at the end of the day, if you want to succeed, you're right: the goal is succeeding.

Yeah. And in “play well together,” and in an NBA team, you're going to know in 2 hours. You're going to know.

Professional sports is maybe the last bastion where there's no room for the “I'm okay, you're okay. Let's talk this out.” “Oh, your team's talked too much.” It's like extreme accountability.

Extreme accountability, extreme teamwork. So, I learned some things that would have been very helpful for me to understand at Microsoft. I'll give you another one: reference checking. Everybody does reference checking, right? How good is the reference checking in most businesses?

Not good.

Well, most people call front-channel references, which has never made sense to me. Or you call somebody who probably doesn't feel like they can give you an honest answer because they don't want to get sued. In basketball, you should see the amount of reference material we have on a guy before we draft him. People have talked to their old coaches, they've talked to their teammates.

Right.

And it's not just—I mean, that's kind of what scouts do. They've watched them play. They've been to practices. They kind of know what they've talked to references about: work ethic.

Imagine if you could scout your future employees. You could just go hang out at their current job.

Yeah. Or you could talk to their parents. I mean, there are so many things. The draft choices are such a crucial decision, right? Because free agency—I mean, you mostly know what you're getting, right? You have a body of work. You can see the body of work. You may know what happens behind the scenes; you may not, right? So there's some risk to it, but there's a body of work.

A draft, you get 2 choices every year.

Well, we traded some away, but yes. Right, right, right. But in theory, yes, David, in aggregate, every team gets 2 choices every year, and you could choose to deal those choices. So that's part of it.

Yeah, yeah, but that's hugely, hugely impactful.

And you're dealing with one other thing. My wife reminds me boys' brains don't fully develop until they're, what, 25, and we're drafting guys who are 19, 20, 21. So you're also having to say, by everything I know, what do I project that guy looks like as they get into their—you could say you enter your prime around 27. What do you start looking at, though? You're going to look pretty good or not by 23, 24, 25. So you have to sort of have a progression of what you think happens to the young man when you draft him.

And so reference checking is a bigger, much bigger deal, I found. People say, “Ah, well, it's simple. It's sports.” The strategy decks I get are 35 PowerPoints, 40 PowerPoints, easily, to go through: “Okay, here's our strategies. What about this? What if? What about this? What do we do here?” It's complex. We have a PhD physicist who is a key part of our analytics group and focuses on our analytic systems. It's not like this stuff's not complicated. It is.

Analytics has become this really big buzzword in sports. Where do you see real alpha actually happening in data science and sports versus what's just table stakes at this point?

There are 2 ways to use analytics. One is for game planning: literally, what does this tell us about the best way to guard Anthony Edwards in this situation or these situations? It's very helpful for that. I'd say the data is probably table stakes, honestly. The way you use it, not so much. Do you ask the right questions? Maybe not. Does the coach really understand and embrace it? Are the analytics people really able to mind-meld with the coach so that coaches get the insights they can for game planning?

The second is what about drafting and trading? Analytics are actually a little less important in that instance because they don't really tell you how it's different if you mix Charlie with Harry than if you mix Charlie with Bobby, and Charlie and Harry haven't played together before. So it's a little different. They are helpful. We have analytics, for example, on all the kids we're going to draft—less valuable than on pros because you're playing against a different level of competition.

Do people have differentiated data?

Not much. The same cameras in the ceiling are recording the same games. Most of the analytics data now gets processed through standard software packages that get licensed to everybody. There's a company called Hawk-Eye, and Second Spectrum, and basically they've built machine-learning layers on top of the raw motion data, et cetera. Every team winds up with the same tools. That doesn't mean you don't need smart guys. It doesn't mean you don't do analysis on top of it.

Has anyone had a breakthrough form of measurement? Is there an example in the last 5 years of a team that's had a great data source emerge—a different data source than other people have?

No, I don't think so at all. What people emphasize in terms of what they look at could be different. I think it varies very much by team. There are teams at the draft who just have you take a psychological test. You get to interview a set of kids, and they might just have you take a test. Other teams, it's all about the interview. Some people—I don't know if they have them see psychologists. I don't know. But people will use different techniques to try to do some of that. It's a little different than analytics, but it gives you the sense of how you assess what's important.

Fascinating. Interesting. How does Intuit Dome fit into all this?

I love Intuit Dome. We talked a lot about products, and I've been involved in, I'll say, the visioning: What should this product look like? And particularly a number of them—both Windows, but also certainly back-end products, meaning they're not customer-visible. But I would say Intuit Dome is probably the product for which I had the clearest vision I've ever had. I knew what I wanted. It evolved some because we went and looked at a bunch of other arenas, but I had a point of view. I know what user I wanted to make happy.

I bet a lot of people aren't familiar. What is the thesis behind Intuit Dome?

I wanted to make Intuit Dome the best place for the hardcore basketball fan and particularly the hardcore Clippers fan, right?

Sure, sure, sure. Of course, because we're the only team that plays there.

You've got another team that plays there every night: your visiting team.

Yeah, we're not trying to help them.

You're not trying to help them. So, yes, Clipper fans. But I wanted to—we're going to have the Olympics. We'll have every Olympic basketball game at Intuit Dome. I want it to be great for those environments. We have some college games or high school games in there. Basketball, basketball, basketball.

So you sit in there and you're a fan. You want it to be a live event. It's got to have energy. It's got to have intensity. If you're a basketball fan, come on, let's go. And so you want it tight. You want to have it reverberating with people who are cheering.

We built essentially a whole side of the building structured more like a college gym: long and steep. There are no suites on the side. We even built a student section right in the middle. It's standing room only. You must stand. That's what you have to agree to if you're going to sit there or be there. You have to agree to stand. You have to agree to cheer. And if you don't, we'll find you another place in the building to sit. But you can't wear visiting gear or paraphernalia on that whole side. 4,000 seats. We'll move you otherwise.

Wow. It's small.

The number of seats is a little small, but the way we pulled it together is not. There's no hockey. I didn't want hockey—not that hockey's not a great sport, but you have to spread it out. You have to spread people out because the rink is bigger than the court. Very different. Basketball. We put in this—we have an acre of scoreboard.

Yeah.

Okay. Yeah, the Halo Board is unreal. More statistics.

More statistics. We went 4K from the start. I didn't realize it's an acre. You have an acre between the inside and the outside. It's almost an acre. It's the largest indoor screen in the world.

Yeah. For sure. And what you were describing before is the Wall. For any fans—or listeners—who haven't seen a game there or seen any of the interviews you've done about this, it's an unbroken 51 rows, all the way up. I call it the student section. We call it the Swell—Clippers waves. Get it? Swell.

The Swell is right in the middle. They do a chant before the game starts. They're chanting, they're making noise, a bunch of them. They'll find weird things they want to bring to games—funny posters—but you basically sign up first come, first served. If you're not there early, you're not in the Swell that game.

So, we oversell the section. It's $1,000 for the year, which is only $25 a game. Hell of a price, but you're expected to go to an NBA game for $25. You're expected to deliver the goods. You have to bring the value.

And the thesis behind the Wall, if I'm understanding correctly, is it should be easy to be a Clipper player, but hard to be an opposing player.

You got to bring the value. We put it right on top of the visitor side, so it makes noise right into the visitor's huddle. We put the Swell right behind the backboard, so basically when you're shooting free throws on that end, you're looking right at the Swell.

And it makes a difference. I saw data that said the lowest free-throw shooting percentage in the league for the visiting team was against the Wall. Like, Steve, this worked.

That's what I wanted.

What do the other owners think about this?

We've had a bunch of people come through and look at the building. Would I be surprised if a number of the new arenas that get built don't have a Wall? No.

But at least you've got a duration of your advantage, because not every other team can build or remodel an arena.

But you also have to remember, I took some financial hits on this. We have fewer suites, less revenue, and we only charge $1,000 for a season ticket that gets you pretty close to the damn floor.

And you didn't have any public funding for it.

In terms of the cash, California—you can't have public funding for arenas. That's why we don't.

So, you paid for the whole arena, and you're going to have a slower payback on that because you have less revenue opportunity.

Yeah, we took a revenue hit. Definitely, we could have made more revenue on that side if we had done things a little bit differently. But it's about basketball.

We have a lot of toilets—3 times the average or something like that.

Why?

It's about basketball. Get out and get back into your seat. Don't miss the action.

We started out with a lot more concession stands, and then we said, “No, no, let's just do this completely frictionless.” So, if you register your face, you just walk in, grab what you want, and leave. If not, you could just tap your phone on the way in, grab your stuff, and leave. There's no checkout.

We don't serve eclectic food—a little bit of everything, the same thing everywhere. Why? We don't want you walking around having to look for your favorite food. No, you're going to get the same great stuff everywhere.

It turns out 85% of what gets bought is in 5 items anyway. It's a hamburger, a hot dog, nachos, chicken tenders, and—I don't know, I'm not remembering off the top of my head.

Is part of the calculus of this for attracting players, too?

Sure.

But I'm thinking even if you know your opponent's going to have a lower free-throw shooting percentage in your home arena.

Sure. I think players have said they think it's really cool now, and that's good. That's good.

Players' facilities are also good—i.e., the training facility. I mean, the training area, the practice area, our outdoor pool and sauna and cold plunge, our weight room, our sports performance center. That stuff's all, I would say, pretty good as well.

Very good.

So, we've done a bunch of things. We have the best refs' room, I think, in the league. We called the refs' union and said, “What do you guys need?”

The media area—we said, “Look, if we're going to build a new arena, our visiting locker rooms should be the best in the league.”

Best weight room? That's your sales pitch to visiting stars?

Exactly. Yeah. We say, “Hey, we care.” And we care about everybody, and then we make it about the basketball, in and out.

I mean, we have public art, as required, some of which is basketball-oriented. But our major piece of public art is a Clipper ship whose masts are replicas of basketball backboards from around the world.

Basketball. Basketball. Basketball.

Our art inside the building—we have a high school basketball jersey from every high school in the state of California. It looks like art, almost, because they're nice colors on the wall. Basketball, man. It's about the basketball.

This building feels like your personality turned into a physical structure—the competitiveness, the loyalty, the fixation on what matters to the customer.

Yes. Yeah, it is. Look, I knew—you know how oftentimes startups come about because the founder is in love with some topic and builds the product they wanted to use. I think that happens a lot.

I don't think people start by looking at the market. They say, “Ah, I think...” You know, I think Zuckerberg did that. Bill Gates did that. Programming—everybody does it, right? That's what you did.

I didn't. I didn't try to go out and survey. We could have designed for the—let me call it—the contemporary audience. We would have had more lounge space. We could have designed for what I'd call traditional, long-term fans. That's kind of how I think about it.

We could have designed in a lot of ways. I designed for me, in some large measure. And it turns out Clipper fans are a little bit like me, because some of them are long-suffering. The team wasn't good there for a number of years. People are die-hard. They'll come up to you and say, “I'm 89,” which really means they bought their season tickets in 1989 and they've been there.

Now we've exploded in the last—whatever—14 years. We haven't had a losing season. Hopefully, a championship here at some point.

Yeah. When are you going to overtake the Lakers?

You know, there are battles in tech where you just have to be patient and long-term. Our goal in L.A.—it's weird to have a town with 2 teams—is to be long-term grinders at that. And we want to beat them every time on the court.

It's okay to have 2 popular teams. Los Angeles County, for gosh sakes, has pretty much the same number of people as the state of Ohio. So, there's plenty of people to be fans.

We don't want to be, quote, “little brother.” We don't want to be the team with a nice 20% market share. No, we want to get our fair share. We're never going to get 100%. The Lakers have tradition.

So, just like at Microsoft: patient, long-term, hardcore approach. And if we don't do that, no—the Lakers have their position. They've earned it. They've got a lot of championships. That doesn't mean we're not going to keep coming and coming.

Steve, thank you so much.

Thanks, man. Thanks, David.

Thanks, Steve. Appreciate it.

Woo, David, that was fun.

Yes, it was. I've always wanted to interview Steve Ballmer. In fact, when I was at Microsoft, I wasn't a podcaster then, but at the time I was such a junior employee. There was a complicated landscape that Steve was navigating between the product set, developer relevance, the shifting landscape underneath him, and whether Windows would be the interesting bet to make going forward—personnel stuff, board stuff, eventually CEO transition. That is not a job I want.

It's kind of fun for us as a show, too. Obviously, this is meaningful for you personally, but when we started the show in Seattle in 2015, Microsoft—the Microsoft transition, Steve, Satya Nadella—this is what was in the water. This is what we all talked about at Madrona and in the Seattle tech ecosystem.

And it's not clear that Microsoft was going to be this amazing juggernaut that it turned into. Obviously, Steve had planted some seeds in enterprise and what would become the juggernaut of Azure, but we were early in Satya's tenure when we started the show. Everyone had high hopes. He had started to transform the culture, but it's come a long way.

Steve knew how great Azure was going to be, but the rest of the world didn't yet.

Yep. So fun.