Gates: Why is Windows so cheap?

This week, Microsoft CEO Bill Gates hit San Francisco for the official launch of Internet Explorer 4.0 and sat down with editors at CNET's headquarters to talk about software, standards, and market strategies.

Does the AOL purchase of CompuServe change your goals with MSN?
Oh, they don't make much money. It's not like some model of "Oh, hey, you just do the data--you'll make a lot of money! Now, how much?" No, I think their purchasing CompuServe is interesting. It gives them the big number. It makes us No. 2 clearly. They're a very strong No. 1 in terms of just numbers...I would say it changes our strategy.

All things Microsoft: CEO Bill Gates in a CNET interview.

But they haven't made a dime in 12 years or as long as they've been in business. How long will you go investing in MSN and some of these online ventures before you pull the plug?
Well, I have a longer time horizon than most other players in the market. So as long as I think that the net present value of the activity is a significant positive number, the fact that the near-term years are negative numbers doesn't bother me. I'm going to be here in ten years. But if I decide that the net present value summed all over those years is never going to work out, then hey, why do the thing?

I have two measures for the people who do interactive stuff: Are you No. 1 or No. 2 and if you're not, then I don't think No. 3 or No. 4, no matter how many people are there and all, I don't think they're going to do that well. And is your category developing enough that it's going to be a significant category? So some of the things that we're doing in that area I'm sure will be quite successful and then some of them may not be.

Will it take you ten years do you think?
I don't see any of them that will take ten years at this point.

When do you think that technology for WebTV will be acceptable, that a lot more consumers will go out and buy it? Is it a year?
I think we're going to have a big Christmas this year. But when you talk to the people who use WebTV, they're very enthused. Or when you talk to people...say I've got a relative, all they want to do is email and I'm going to get them set up on WebTV. There's this very strong reaction to the product. And so I think it's ripe to do very well. Last year it was a totally new product category. We had to get the distribution going--it wasn't us at the time, it was just WebTV as a small company. Now they've got the Microsoft brand. We're going to put quite a bit more marketing behind it. We've got Sony and Philips and now Mitsubishi will have a product this year.

So I can virtually guarantee we'll do twice as much this Christmas as last year. Will we do four times as much? Five times as much? That's hard to say. It's the word of mouth spreading and how people see the product fit-in. So we think this year there will be a big uptick and then the next year after that...It's not just at Christmas, but it will share the same kind of seasonality that home PCs do or a little bit greater.

Do you see Java to be a bigger competitive threat than Navigator/Communicator?
Well, we're the most successful provider of Java tools in the marketplace. Visual J++ is doing extremely well. In terms of performance, debugging, rich development environment...I feel like our Java guys have done a great job in the marketplaces responding to that.

So Java, the language, we see as an important language. In no sense is that a competitor. Now does it mean C is going to go away? We don't think so. In fact when I was at the PDC [Professional Developers Conference] I said, "Oh, I think C for the group there, that C will continue to be the most popular language." And there's this huge applause. It's like, "OK. Hope you love C. Go ahead, stick with it." So it will be an important language: VB, C, COBOL, and then there's some like PowerSoft, ABAP, that will be there. Delphi as well.

In terms of the notion that people are going to write applications that don't take advantage of an operating system--that's where you see a difference in opinion. When I buy an application, I say "Hey, does it use my security model? Does it use my user interface model? Does it use the clipboard? Does it integrate in with the other applications? Does it use what this platform is fastest [at], does it make the graphics calls that this platform is optimized for?"

So as a buyer of an application that somebody says to me, "No, I'm not taking advantage of anything in your operating system," I'm not sure how strong of a pitch that is. But if they say "OK, well I've got this duplicate operating system I layer on top of that"...Just think Macintosh. For Microsoft, how strong of a pitch was that for the old Mac user to say, "By the way, this application adds nothing specific to the Mac--no balloon help, no color management, none of the user interface guidelines. This application is for me, the developer. I was lazy, and because I wanted it to run on a wristwatch, sure I only used 100K of memory and I only used this much of this much of the screen because, hey, I wanted it to run on a wristwatch, so I don't use this thing."

There's a question of will people write applications that actually use the rich environments: storage, scheduling, graphics--you name it--or won't they. We think they will, whether it's our environment or Apple's environment. Now they can do that in Java and other languages. I wouldn't say it's a big competitor. That's just a question of "Do users want exploited applications?" It's not a competitor. We compete with Solaris now--their volume versus Windows' volume or even just NT volume. It's not huge, but they do sell into a part of the market that we are doing better and better in all the time. And we're quite focused on "OK, what is it that we need to win even more of that business?"

Would you clarify the controversy over the Java licensing. Is it going to end up in court, as Sun says?
Well, Sun took an unusual position because they're sort of not only the brand-name holder of Java, but they're also a competitor of ours in terms of selling operating systems. We've done very good Java implementations. I mean everybody can just look at them--run the test cases, see how they work, see what the speed is. That's the objective data.

The fact that Sun is afraid of us--just go to any of their speeches, they talk about Microsoft. Sometimes they forget to mention Sun. So is Sun thinking about Microsoft? Apparently. Are they thinking friendly thoughts? It doesn't sound like it. You have to ask them. But it's not sounding too friendly. Now does this mean they're feeling some pressure on overpriced workstations? Potentially.

And it's tough, it's tough to compete with a PC model when you say, "We're open," but then it turns out that as soon as you buy one of the Solaris systems, it's very hard to buy hardware from somebody else. Whereas in the PC world, you buy hardware from Compaq today, HP tomorrow, and IBM the next day. Which is open? Which gives you the most choices of applications, which gives you the most choices of peripherals, which seems to be the most competitive in terms of basic system pricing?

And so Sun is in the model of the classical computer companies where the hardware and software are tied together. And we don't represent that model. We're part of the PC model, which is Compaq, Gateway, Dell. Many, many companies are playing in the PC model, which has become a huge part of the computer industry, the majority of it.

What exactly is the issue when they come out and they made these public pronouncements that Microsoft wasn't following the licensing agreement?
Well, I think it's very consistent with everything else they say. They appear to think we're a competitor...and they want to say anything they can. Certainly we're following the license. And if they honestly don't believe we're following the license, at some point they'll describe that to us.

It was reported that Sun is going to examine IE 4 code and if there are inconsistencies, they might look into taking this to court?
When we did the beta of IE 4, they had a certain set of test cases and when we do the beta we said "OK," and this is what we have agreed, is when we did the beta, there were a certain set of test cases--we passed all those test cases. Now they added test cases have nothing to do with running Java applications subsequently. And I know that [Microsoft group vice president of applications and platforms Paul] Maritz is talking to them about that. We passed the test better than anybody. I mean, just go look out there, run the test cases. And ask yourself why are they raising some that doesn't really have anything to do with compatibility with Java applications. By every objective measure we've done a very good job on Java and if they're going to attack us, fine. I think that's what they seem to do on Monday through Friday and on weekends, if they're working. So we'll put up with that. We're not attacking back.

What do your customers say? Are they confused? Are they sitting on the fence saying "Geez, here we go again!"?
What fence would they be sitting on?

Do they say "Geez, let me just see where and how this thing plays out before I devote 20 programmers to do just Java-only applications?"
Well, most people are buying packaged applications and modifying those or working on applications they already have. So what's the most popular corporate development language? VB by a factor of three over the next, which is C, which is a primary tool that is far more popular than the next, which is COBOL.

So there are a lot of different things people do in development, most of which is incremental improvement. If what you're saying is that Sun is screwing themselves, that to the degree they make these complaints, it makes people concerned about using Java, then you should mention that to Sun. We're not making any complaints. So if they want to scare the people, I guess they must enjoy doing it. It certainly gets their names in the paper.

Oracle has often said that SQL Server doesn't represent a huge threat to them. I'm wondering about your view as Sphinx [the next version of SQL Server] gets ready to roll out, and how that might play out in terms of Oracle 8?
I think they underestimated, so don't tell them. If they think it's not a threat, they should take the rest of the day off and the rest of the month too.

How do you close the gap?
Well, there's a general approach. This [Sphinx] started about three years [ago]. You start hiring very good people. You say "Hmm, who should we hire?" Look at our database team and look at the amount we've put into database R&D and look at Sphinx betas. It's an objective world and maybe Oracle is right--we don't know how to write software. I'm not going to clue them in. But there are people who are working long hours building an amazing database product.

Do you think enterprise customers want to buy a package of BackOffice-bundled applications that includes a database? Or do you think they may still want sort of a best-of-breed approach with an Oracle style?
Look at Microsoft Office. How did we win? Well, we won every spreadsheet review with Excel. We won every word processing review with Word. We won most presentation reviews--not all--with PowerPoint. We did best-of-breed products, then we integrated them together and created the Office suite. But we won in each standalone category, we won before the integration gave us further customer benefit. And likewise in BackOffice today, SNA Server competes against ComServer from IBM. That product wins all the reviews. Exchange competes with Notes. That product wins some of the reviews. We're doing great work, we're very serious there.

SQL Server: Depending on what people use it for, it does very well in terms of simplicity of setup and the admin and the programming model. It is best-of-breed in terms of some of the very high-end benchmarks. We've said Sphinx is where we have a chance to match or perhaps even lead in those areas that we don't today. See, we're pretty realistic. And the thing about Microsoft is we hire other more-smart people. When we get into a field, we're pretty good about finding out who the best people are, whether it's [Microsoft Senior Researcher] Jim Gray or on down through the team.

I feel we did what we always do very well. It takes time. As I say, this thing started three years ago. Over the next few years we'll just get better and better and better. Sphinx, which is the code name for SQL 7.0, is a milestone. It's not the end of the journey. The day we ship SQL 7.0, Oracle will dominate the database business. A year after we ship SQL 7.0, Oracle will have the majority share in the database business. That's easy to say. Do they have any competitors? Well, Larry [Ellison] has decided that he can ridicule Informix and scare customers to not buy them. He's good at ridiculing IBM. He just doesn't think he has any competition. That's very different from my attitude. And I wake up every day and say, "What are our weaknesses?" I see competition everywhere; he sees it nowhere. It's just a different personality.

Would you consider buying Informix? Do you think their technology is good enough to put you at par with Oracle 8.0 today?
Well, we've made our decision on how to do databases; and that's hiring the team; and we actually started with a Sybase code base. But we've hired our team, we picked our code base--that is our database strategy. So it wouldn't make sense for us to buy another database company. Informix has very good database technology as a company, but our strategy is exactly what we've done. It is focused on exploiting NT in the right way and built around the programming models that we're putting forward.

With Steelhead and Ross, NT has had some interesting networking features added to it. We're curious, given the history of Microsoft on the desktop, how it's added more and more functionality there--PC server-side, a similar path for NT on the server side--adding more and more protocol support, etc., to allow a richer internetworking NT network experience?
Well, you look at Steelhead, [and] it's a pretty amazing product and certainly we have had a very good reaction to it. Our goal in doing communications software is to make NT a great end node, which sometimes in a branch, it might be the end node and you connect NT up to the network or a server in a corporate network. And so supporting...we have added a lot more protocols. And we'll keep adding things there. Our goal is not to be an intermediate node in the network, although some people actually license NT from us and use it in products like that. But that would be companies that are focused in that area.

Do you see a large market in that space? OEM the internetworking vendors?
We're certainly talking to everybody you can name, starting with Cisco and on down about do they see NT running in some of their products. And we're willing to do special versions and maybe we have to do special licensing for that. It's not a market NT has been in, [though] there's a little bit of it right now for some of the RAS dial-up servers that are NT-based. And we're seeing a reasonably high level of interest. As a percentage of the total NT volume, no, it won't be overwhelming. I mean desktops...that's where your numbers are is desktops and portable systems. So if it was a substantial deviation for the NT team, we wouldn't do it. But it turns out it's a general purpose operating system, and a lot of the rich features there are helpful as a platform. But all the value-added pieces that allow you to do intermediate communications-type products, it'll be other people who are building those on top.

Technology is evolving very quickly, especially on the Web. But with Java, among other ways to deliver content, how do you develop standards and not be taken for a bully? And who should own the standards?
Well, the word "standard" means that no one owns it. Well OK, let's differentiate two types of standards. There are standards in the official sense where no one company controls the brand, controls the evolution, anybody can participate or make suggestions. The IETF [Internet Engineering Task Force] and W3C [World Wide Web Consortium] are wonderful organizations that do an amazing job of making that process work. And we've put a lot of man-hours into it and we think those are great contributors to the industry. Those are open standards.

Don't you see W3C as too slow?
Actually compared to all other standards bodies, they are impressively fast, honestly. IETF and W3C are pretty fast. Now, when you're a company and you're doing a product, can you implement a cool product that's going to win and only stick to standards? We support all of those standards--every HTML thing--we've been very, very careful about all of that stuff. Sometimes companies do take things that they are proposing or other people are proposing and put those in the product. And there needs to be a word for that, that yes, we've made it available to the standards committee and we think there's a good chance that they'll adopt it, but it has not yet. Vs. something that you just do on your own and you're not submitting to a standards body. Now all of those are legitimate business strategies. It's just people have to be clear about what they're doing with those things.

We've certainly been a company that's given Sun a hard time about trying to have their cake and eat it too; that is, own the Java brand so they can tell me, "Don't ever say Java, don't label things that way." Only they can propose extensions, they're the only company that can do anything about the thing and still try to pretend that that's a classic standard that's really open to everybody's innovation and contribution and managing the test case or the rhetoric around the test case in a way that's neutral to all participants.

So you distinguish that you're not opening up Win32 APIs from what Sun is doing because you don't promote it as an open standard?
Well, "open" is one of the most abused words in the world. Remember, Unix is open and it locks you in more than any other environment. So the word open is a tricky word. Everything seems to improve when you put the word open in front of it. True open now, because open has been abused so much, you've got to stick "true" in front of "open" now.

Is Win32 open? Hey, go down to the bookstore and count the number of books that you can buy that document the Win32 interfaces. How much do you have to pay me when you do a Win32 application? When do you have to tell me? You never tell me, you never pay me. What about Nintendo? They decide when you can do it. They charge you for the thing. So there's many dimensions...I don't pretend that when I do a new version of Windows that everybody gets an equal vote. The Windows is a trademark of Microsoft. It is not a standard in the W3C sense, but it is open in many senses of the word.

Is it OK if desktop software operating systems are one thing [and] the Web is another thing because of the pervasive nature of it and the content is going everywhere. Is it OK on a Web browser to use things that are not standard?
That haven't gone through the full?...It's up to buyers to decide. There's a lot of things in a Web browser, like the user interface, that never will be standardized. So is it OK to innovate the user interface of a Web browser?

Only in free countries is that allowed. How about making a Web browser faster? Oh my god, now this is a problem! So yes, innovation is still OK. We do have things like VHTML that are not yet official standards. I don't know of any browser out there that restricts itself to things that are official standards. Maybe IE 1.0...we should go back and look. And it might be the only browser that does that because Netscape, all their old versions had that layered thing that was rejected. So they don't meet that test.

Let's give you a scenario. DOJ [The Department of Justice] decides that bundling IE with Windows is anticompetitive and orders you to unbundle. What's the game plan? We assume you guys have thought this through, that at some point DOJ could make us eat it.
It's not a hypothetical that I'm going to respond to. The basic idea of...you know, operating systems, we're going to keep innovating in operating systems. We spend a billion dollars a year just innovating on operating systems. And people are asking for a lot of stuff. Take TCP/IP stacks--they didn't used to be part of the operating system. You used to go out and pay money. And there were 11 companies that were in the business of selling you TCP/IP stacks. Now this was ugly. Some interoperated with some, they hadn't done DHCP [Dynamic Host Configuration Protocol], it wasn't optimized. So what did we do? We listened to our customers and they said "This is screwy. Put a good TCP/IP stack in the operating system." We did, we integrated that in. That's called innovation.

Now those companies, some of them took their good engineering and chose to do value-added products on top of that or to go into new product areas. That's what happened in the marketplace. And all these issues are just well-known. There's no uncertainty of any kind about whether you're allowed to innovate and put new features in operating systems. There are cases over 20 years old about IBM innovating in its operating systems and being told "Yes, you are allowed to put more things into the operating system." And it's not some narrow definition. Did we put Notepad in the operating system? Yes, we did. Did we put Paint in the operating system? Yes, we did. We create new versions of Windows and we decide what's in those products. And we have total latitude of what we want to do in terms of innovative products there.

In this case, users said to us, "We want to get onto the Internet. We not only want to browse, we want to do email, we want to do whiteboard sharing, collaboration...we want to do that in a simple way where we just buy a Windows PC and it's there. And we know that it works and Microsoft provides all of its support services for us." We heard customers asked for that. We delivered to that.

If somebody said in the future you're not allowed to innovate in operating systems, that you're not allowed to put things in customers ask for, I think that will be a sad day. But not just for Microsoft. I think it will be a sad day for technology, and it would be like if they told the car companies, "No, you can't put headlights on because, hey, there's a guy who wants to sell headlights as a separate thing. Radios--you can't put those in."

What about Office to Windows? People are asking...
If I could put Office in and sell it for exactly the same price I sell it in today? I'm perfectly allowed to do that. Now that, to me as a businessman, sounds like a way of throwing away $5 billion a year. And so I personally haven't gotten too excited about it, but my lawyers would have absolutely no problem with it. It would be economically dumb.

Now, if I try to raise the price of Windows, then that just makes it easier for people to compete with me in operating systems. The question you should be asking yourself is why do we keep the price of Windows so low? That's what you have to ask yourself. Only when you understand that will you understand Microsoft.

But on the issue of OS, you have over 80 percent of the market?
Why do we keep the price so low? Think about it.

Photo by Sara Gummere, CNET