Remarks by Bill Gates

Synergy '99

March 30, 1999

BILL GATES: Good morning. It's great to have you all here, and have a chance to talk about the opportunities we all have in the software business. I think there's never been a more exciting time. The pace of innovation is faster than ever before. And the recognition that the work we do will be key to how companies operate is at an all-time high.

I spent last week traveling around, introducing the latest book I wrote, which is called Business @ the Speed of Thought, and it's about this idea that the whole way that companies make decisions, the way that information moves inside their company will be very, very different, and companies that seize that opportunity will get ahead, and the ones that don't will be passed by.

Of course, the idea that you can use the Internet for e-commerce, the idea that it allows you to reach out and partner in new ways, there are so many good examples of that, and that's really what the focus of the book is, to share those examples, talk about the lessons learned, and write a report card so people can see how far along they are in taking advantage of this approach.

The interesting thing is that from a hardware point of view, it doesn't require that much more than what people are already doing. They're buying PCs and networking them together. What they're not doing is buying the right software, which is where we all come in, and they're not stepping back and thinking about how it really drives the processes and the information related to those things.

So, in the next few years, I think we're going to see even greater levels of investment, and a clearer sense in people of what the value they get out of having this digital approach is.

In writing the book, to really explain the idea that there is a new kind of behavior here, a new way of thinking, I had to invent some new terms. For the consumer who's thinking of the Web as something that they use many times a day without that being a special activity, I talk about the Web lifestyle.

Today in the United States, less than 5 percent of people are doing this. There's about 30 percent of the people who say that they do something online once a month. But for most of those people, it's very light use; a little bit of email, maybe going out to one or two Web sites. And you find the people who are living this Web lifestyle mostly around universities or high technology companies. So, all of us are probably a lot more immersed than the population as a whole.

Even though it's only about 5 percent in the United States, that's a much higher number than you find anywhere else, even in other developed areas of the world, like Europe, the number would be well under 2 percent. And I think part of the efficiency of our economy, the way we're getting ahead, is because we're using technology in a deeper way, and spreading that out to the businesses to take advantage of.

When we talk about business and somebody at their desk doing a knowledge worker job, my idea is that we should raise people's expectations. They should expect not to have paper forms. They should expect to have sales information in a pivot table format where they can dive into it and see it in rich new ways. They should expect to be able to navigate the information from all their software systems in a way that allows them to come up with new ideas and share things with other knowledge workers, whether they're inside or outside the company.

This is not just taking a screen and making them work through a structured process. It's allowing the information to be there for them to plan and do things in a better way than has been possible before. And so, it really requires thinking about the productivity tools and workflow enhancements that are being done in products like Office, how those come together with the vertical packages which hold a lot of this key information.

And it's only by having that right combination that we'll be able to achieve these scenarios that I think are critical and that I talk about in the book. And so, these workers who really drive this, and have these high expectations, they're part of the Web workstyle. And the organizations that are taking this approach, they're building the Digital Nervous System.

Just yesterday, Microsoft announced some of the things we're doing to structure ourselves for this next decade. We see it as a decade of unbelievable opportunity. The key things that the company decided to bet on when we were founded, the idea that software would be important and you could build a company just focusing on that, and that the PC would become this incredible tool, both at work and at home, those have come true even more than anyone would have expected.

But now, as we look at this, we see so much opportunity. We've got to get more parallelism in the things we do. We see that we're going to have a lot of devices. The PC, the full screen device, is the primary one. The full screen device is where you're create documents, edit documents, annotate documents. And that will never go away. That is a central thing that you use.

But you'll also have smaller screen devices that are complementary and you want to see your information on those. You'll have lots of things that you're doing with the devices in the home, the TV and the PCs there, that wouldn't have been possible in the past. And so, we've really structured ourselves to go through and have five divisions, each of which is tackling some of the key challenges.

We're going to have parallel action, we've aligned these around customers, and it's a long-term approach that we're taking. The specific divisions that we've built, and I'll just go through those real quickly. We've got the Windows division that's focused on the IT challenges. That's Jim Allchin's group. The big thing there that I'll talk about a little bit this morning is Windows 2000. You heard from Brian Valentine who runs the development in that group earlier this morning. They're competitors are high-end UNIX, Linux, Java OS, there's quite a bit of competition in that space.

We've got a group that's after business productivity, that's where Office and Exchange are. And that's run by Bob Muglia. We've got a group going after what we do for developers, and that's Paul Maritz focusing on the incredible expansion we're doing with the applications server and COM+.

Then we have Consumer and Commerce, which is the online activity, MSN and all our e-commerce things.

And then finally, a group that's focused on Consumer Windows and the scenarios around that, that division also works for Jim.

It's five different groups, each with a key set of scenarios, each with marketing teams within those groups.

So, we see our role as being the same as it's always been. Providing very powerful, high volume building blocks, taking an approach that they run across all the latest and greatest hardware, and driving the standards that allow these things to work not only with each other, but with all the other systems that are out there.

This is a fascinating period in terms of advancing standards. We are really driving XML as a format to express semantic information. And so we see that in every vertical area the kind of work we have done to define object standards, OLE-based standards, we'll support those in an XML format and we'll even take some of the industry standards that were not based on our technology, things like EDI, and move those to XML. And then we'll optimize all of our tools and our databases around XML, because the kind of rich interchange that we expect is not just simple records, it's going to be these graph-structured XML-type objects that are of great importance. The defining schema, defining schema exchange, defining the tools for all this, that's a very top priority. And, in fact, our developer division is the place where that all really comes together in a big fashion.

We decided that our key products looking forward will continue to be Office, BackOffice and Windows. There's a lot of products around those, things like Visual Studio, that are of critical importance, but we’re really just enabling better scenarios in the main products. Part of the strategy has been that instead of having extra cost add-ons, we're going to build the rich functionality into Windows.

So, for example, when we did transaction management, it would have been easy to say, IBM charges a lot for that, it's a very complex thing to develop, why don't we have that as a new P&L. But, our view was that if you want to make it pervasive, get everybody to take advantage of it, and to support the kind of scaling we want where you really need transactions, because if the hardware messes up or the software messes up, the ability to shut down part of the cluster and yet turn the work over seamlessly to the other systems, that requires transactional boundaries. So, in order to make that pervasive, we decided to build it down into Windows itself.

Now, in some ways, by putting all the rich functionality in Windows, the queuing, the message-based approaches, the transactions, we haven't given the visibility that they really deserve. Paul Maritz and his developer group will make sure that the message about how all those pieces work together, how the tools make that very easy to take advantage of them, that that message gets out even better than it has to date.

Built into Windows is pretty much everything you need with the security, the directory, and avoiding your having to duplicate the systems-level functions, that's really a key goal for us, because we don't want to have people have to duplicate each other's efforts. We want the platform to do as much as it can.

What's probably most interesting are the initiatives we have across all our products. Those of you who have been here the last couple of years have heard about the Internet and what a center role that's taken on for us. In fact, some of the standards work we do, where we've got more people on Internet committees now than anyone, the standards work we do here, we think, is very critical.

It's not just XML, although I put that at the top of the list, it's also the quality of service guarantees across the network, so we can start to have video and audio flowing across that network without crowding out the other traffic, which is often a higher priority. Corporate administrators aren't even letting people do the NetMeeting screen sharing we've got built into Windows because we don't have that traffic shape. So, as we work with Cisco, and others, including the standards committees, getting that to be just a common scenario that you expect out of every PC, that is within our grasp if we can get the right pieces to fall into place.

The Internet isn't standing still in any way. Certainly, the Internet today is far more powerful and impactful than it was a year ago, and we see constant evolution there. In fact, the only thing that's going to hold that back is when you get outside of the business realm, and you look at having high speed always-on connections out to consumers. Unfortunately, there's no Moore's law that can magically make that happen at low cost. And so that will be on the critical path. The PC itself will be improving very, very rapidly. And in the business realm, you will take very high speed Internet connections for granted.

Another key initiative for us has been interoperability. We want to give businesses a chance that, even if they keep some of their applications and their databases on large systems, that they can move all their new development down onto the more Internet-centric Windows platform. But to do that, we have to interact with the classic systems, even things like CICS, where we now have as part of our transaction capability, the ability to have an interconnection up with all of the different versions of CICS. We've also got more interoperability in the last year with UNIX, things like the common log-on, some of the add-on packages, and so that's about 30 percent of our R&D budget is around interoperability scenarios.

It started a long time ago with things like the ODBC drivers, and the transport stacks. Now it's moving up the chain to things like data warehousing schemas, transaction integration, to these XML schemas, and how they'll allow businesses to match up the information they want to exchange.

I want to talk a bit about scalability, because I'd say that, of all the things we're going to surprise the world within the next 18 months, it's going to be how scalable the platform is, how much further we're going to go than even the more expensive platforms, and the reliability that goes along with that. That's very central to our strategy. In fact, part of this organization created a group very dedicated to high-end servers, and they'll have some freedom to do things that really address the incredible reliability requirements up in that space.

Finally, simplicity is a big, big goal for us, because as we get these machines to be more powerful, we can actually reduce the number of commands, certainly we can dramatically reduce the number of error messages that people see, and the cryptic nature of those things. And we have some incredible ways to do that, because a lot of the devices are connected up to the Internet. So, at any time, we can take whatever the scenario is, pass the information along, and get the latest information from the Internet to either update the software, provide a diagnostic, provide advice, and help out.

Windows 2000 is by far our most important product. Office 2000 is a wonderful product, the version of Exchange that runs on top of Windows 2000 is important, SQL Server 7 is important, but none of them really are even close to the importance of Windows 2000. And it's taken us long enough to do this product. There are so many advances in it that, you know, to some degree people said, well, gee, when is this thing going to ship and is it going to be as radical when it comes along as the expectations people had? Well, the answer is certainly yes, and we're very close on this, and in fact, I think as we get beta 3 out next month you'll see the excitement build in a pretty dramatic way.

In terms of really taking advantage of our platform, our number one request is to talk about how the new features of Windows 2000, the state mirroring and IntelliMirror, the Active Directory policies, the new simplified deployment for software, the seamless deployment, how do your applications fit into that? That would be the key item for our work with all of you, testing what you have out there, as well as new products that take better advantage of it. So there's an incredible amount of things in this release, and we're very excited about the progress that we're making here.

It's also interesting to see that for the first time as you get a new milestone, this 64 bit milestone, the difference between, you know, when is a mainframe going to have 64-bit, when is high end UNIX going to have 64 bit, and when is the PC space going to have it, the difference is very, very small. In fact, the critical path in this case is certainly the Merced chip. The Alpha chip is out now, and you'll see not only the VLM support, but true 64 bit support on that, even earlier than on Merced. As soon as the Merced chip comes out, we'll have the full support of that. We have some great tools that are out there, and that make it easier to make the transition from 32 to 64 than it was from 16 to 32, 16 to 32, for all of you who were around for those years, that was pretty messy, because the memory model was so different.

You had segment bases in there, and the whole way you mapped things out was very different. Here we're just talking about taking a linear address base, and going to a linear address base. And so all the tricky issues are deciding which variables and which structures that you've had that are 32 bit are 64 bit, and as you're coercing things back and forth from 32 bit to 64 bit, making sure that works. Well, some of these very advanced tools are good at catching those things, and you'll be able to have a common source code for 32 bit and 64 bit. With Windows NT that's what we're doing. We haven't forked off a separate tree whatsoever. In fact, we have a nice system where anybody who checks in a piece of code that doesn't pass the 64 bit test, they get when they come in in the morning a lot of email, because we run those tests every night, and then we email out every violation that somebody has against that 64 bit test requirement. And so we'll be able to keep 64 bit and 32 bit on track with each other, the same way that we have with things like the Alpha version and the Intel version, as we're doing the new releases.

Now, this new group that we've formed around Paul Maritz to advance our developer platform, the key thing you'll be hearing from them is what we're doing to take COM forward. It's a very messaging-based approach; it gives us flexibility to reach out not only from the Internet to servers, and servers to PCs, but even into small devices. Small devices will be essentially COM-plus objects that you can send messages to, see their properties, and you just program it as though it's an object inside your program, even though you're connecting to it over some type of transport.

Our developer strategy has many unique elements to it. One unique element is that we're supporting all languages. We see a lot of new innovative languages coming along with these tree structured object type data systems, things that are taking XML as a given in the design of the language. Well, we don't think you should ever have to rewrite your code into these new languages. But, you've got to be able to take advantage of them. Java went a certain distance, and these new languages are going much further, and so you'll want to write some of your code in these new approaches.

Then again, there's going to be a lot of code out there from customers that's in COBOL, VB, ABAP, PowerSoft, you name it, there's quite a few things. And mixing and matching is really the only way to let people move forward. So it's an approach that supports all languages. It's a support that we're going to get the COM talk foundation onto many different platforms just like we did with COM. And we connect up to all the different data sources. So the Windows platform has the best tools, but it's interconnected out with the other environments. We want to make it clear that you can do development and create an Internet service, and create code for the corporate server, and code for the PC, and share all that same code. And not have to have different development efforts that aim at each of those different targets. That's our three-tier approach and a key point to innovation for us.

Another initiative I talked about is simplicity. And here, what people expect to be simple is always, whatever is complicated today; as soon as we tackle that, people will ask for more. It used to be complicated to get your TCP drivers set up, it used to be complicated to get your fonts right. Well, those things are reasonably straightforward today, just like device drivers, or the memory management. Mobile computing, having different configurations when you're on the road, that's gotten a lot better. But, what people want today, they want to set up audio-video conferences, they want workflows, they want seamless application deployment, they want to have the ability to never visit the desktop as they do things. There are a lot of demands there, and at every point in the industry. Developers want greater simplicity. End users want simplicity. IT people who see it in terms of the cost, and some of the things that breakdown, they're probably the most vocal, but they're not the only audience who wants those things. So, across the board that's got to be a very key thing.

It's interesting as we think about all of this spreading out to ask, what is the key thing holding it back? You know, at one point people might have said it's the price of the PC. Well, I wouldn't say that that's true today. I'd really say that it's enabling the new scenarios which on the consumer side are things like digital music, digital video editing, putting up a Web site, and making that very straightforward, making it so you rely on that device, you don't have to learn a huge number of commands.

Now, the percentage of people who are doing this, compared to what's possible, is still fairly small today. And we always want to measure what the man on the street is thinking about this, how do these things fit in? And so, from time to time, we go out and do interviews. In fact, we recently had a consultant go out and talk to people, see what are they thinking about the Internet and some of the things around it. So let's go ahead and look at that, and see what they're thinking.

(Video shown.)

MR. GATES: So we have a long way to go, a lot of opportunities to sell new versions, to work on some of the problems there. The most recent releases we've had really reflect the priority we've given to simplicity. Whether it's the dynamic menus, the install on demand self-repair, a lot of automatic behavior that didn't require people to actually learn a lot more to get the benefit. With SQL Server 7 it was the simplicity in terms of automatic tuning, not requiring an administrator in the same way that databases classically have, or the English query, or the ability to use the same database on a cluster at the very high end, down to a laptop for the same application.

Along with the incredible improvement we've had in scalability, we've also been working with key partners, particularly people like Hewlett Packard, to make sure we can deliver great, great uptime. Now we have programs where there's 99.9 percent uptime commitments that are made, and with Windows 2000 we'll take this to a whole other level. The reason is that there are some approaches here, in particular clustering, that allow us to get the best of both worlds, incredible scale and incredible performance. Today, if you want to scale a Web site, because of the software we've got in Windows recently, called the Windows Load Balancing Server, you can go up to any number you want, and have reliability along with it.

At the software level, the work we're doing with clusters, making those easy to set up, making systems management with the directory in Windows 2000, so you can manage all of those very easily, those are very key steps. Also, we're really tightening partnership we have with hardware vendors, making sure that as people like HP move up to 64 bit we've got the pieces in place to fully exploit those systems. We're not only getting the clock speed improvement, but we get, of course, staging more in physical memory, so you don't have to go out to the disk as much.

We also of course get more processors per system. Where you've seen four and eight in the past, these new BUS structures called NUMA, non-uniform memory access, will let us go up to 32 or even 64 processors in a system and still get linear performance out of that. Then you take and combine those together in the cluster and it's all those factors multiplied together, software improvement, chip clock speed, using the extra memory, more processors, more clusters, multiply those together and you can see that achieving levels that no expensive system ever went to before is going to be pretty straightforward.

So we'll really surprise people on the upside here, in terms of Windows-based servers and what role they can play even in, say, an extremely large Web site. And because of that, the development approach people can take really exploiting the platform, they'll never worry about whether they run out of gas on the high end, or whether they can take their application down onto a portable device that people want to use wherever they go.

For us, we see the opportunity as continuing to focus on what we've been good at. Those are the high volume software products, a lot of services to go around that, but the key goal is really driving the R&D, using the customer feedback, and the breakthrough innovations. We've upped our R&D budget over 20 percent a year, it's up a little over $3 billion right now, and what that is very simple. That is software engineers, a lot of them here at our headquarters building these new products. And so we see ourselves moving even faster than ever before.

Now, the model that we started with when the company was founded was that in the areas of applications, chips, building systems and consulting, we were going to work through partnerships. And it may have seemed strange at the time, because the industry had a structure where that wasn't done, it was done within a vertically oriented company. Of course, this model has proven to be the successful one for the PC, and certainly as the PC becomes more of a communications tool, and it's all driven around the Internet, I think this partnership model will be even more important.

Now the most important class of partner for us, by far, are people like yourselves who build software on top of our platform. We've been expanding our programs in that area over time. One that we think has gone super well, and I put it here as a model of how we like to work, is the partnership we've had with a high percentage of you on SQL Server 7.0. This is something that began years before the product came out. It clearly shaped the features we put into the product, because we had a lot of tradeoffs we were making during the design phase. It clearly shaped the way that we optimized and tuned the product, because we were taking the benchmarks that you used to measure yourself by, and applying them here and seeing where you suggested that they way we design SQL 7, or the way that you actually call it, that we could work together to optimize that. We had a lot of you up not only at development conferences, but in the lab, doing these tests, sitting down with the wizards of SQL Server 7. And on the marketing side, now, we've done super well with SQL Server 7, primarily because of that application support.

So we're taking the technical dimension of our relationship with all of you and putting more resources into that, and at the same time we are also putting more resources into the marketing side. So we're out in the field with our ADCU activities, working with all the companies shown here, and really making sure customers understand that the combination of our platform in these applications is the mainstream and something that people can really count on.

And so the bottom line is that we see the advances you're making, and how that ties in with our new work, as the way that we can continue to have a lot of satisfied customers. Software innovation is more important than ever before. Helping people achieve those scenarios is going to make the work we do for customers not just relevant to the CIO, but to everybody in the business, everybody who wants to be empowered by information, everybody in the business who wants to handle customers and reach out to them in a better -- in a better way.

Windows 2000 is going to be the big event of the next 12 months. And that's something where we really look forward to having the same kind of support that we got on SQL Server 7.0. And with the work that we do together, we can make some of these exciting scenarios a fantastic reality.

Thank you.

SPEAKER: Bill's going to take questions and answers now. So, again, we're going to use the microphones in the center here. So, if you don't mind --

MR. GATES: You should bring up the lights, too.

SPEAKER : And bring up the lights. And Bill will answer the questions.

One question, I think, that's on some people's minds that's sometimes hard to ask is, people understand our partnership model, and understand we're investing in the platform and the tools. But, as the platforms and tools get richer, is there a risk that we'll end up competing with some partners, some people in this room in terms of offering solutions, business solutions? People always say, Microsoft will be in the ERP space one day. What's your perspective of that, what would you tell people?

MR. GATES: Well, there are things like ERP where our strategy is very clear. We've chosen not to be in that space. And, therefore, we've had very productive relationships with the people in those spaces. In fact, if you look at the portion of their shipments that use Windows NT and the portion that use SQL Server, you know, those numbers are rising pretty substantially, and we hope to get up to very high numbers in that respect. Certainly, we're putting an incredible amount into that.

If you're somebody who is a little bit closer to the system, for example you're doing rich storage management, things like document management systems, our innovation does change the boundary, and it actually does put down into the system some of the facilities that had to be done as an add-on in the past. We get that a little bit with ERP vendors. They have the question of, should they do their own transaction management stuff, because they're cross-platform, or should they take advantage of ours. You know, that's an engineer-to-engineer kind of thing, where we've got to make sure that we make it attractive to exploit what we've done.

In the document management space, it does take some of what has been visible value-added, and gets it down into the system. So we need to have a close dialogue, like we do to say the virus people or system utility people, who are the closest to the system, and make sure they understand exactly where we're going.

Now, our product cycles, in terms of major innovations, are on the order of two or three years, so the opportunity to see where we're going well in advance, talk that through and see how the value-added can continue to be that much stronger is very, very high. I think we've handled that very well, but it is a dynamic environment in terms of where the system is going.

Our expertise is not doing vertical-type software. So you haven't seen us do a lot of things there. You know, people always say to me, what if IBM bought a lot of these people and did that? And my view is, that would be unfortunate, and it would be a whole new equation. But the fact that there are so many great software companies out there independently pursuing those areas of expertise, that's the rich model that's really making things thrive, and that we're optimized for.

And we have competitors like Oracle, who have taken a different approach, and we think that's great because it means we can work better with everybody here.

SPEAKER: Okay.

Q: What motivates you, and what do you see yourself doing five, ten years from now?

MR. GATES: Well, I think I have a very interesting job. There's no shortage of things to think about and learn about, you know, whether it's these small devices, and how those fit in, and learning about wireless. Making the tablet PC a reality so that at a meeting like this, two or three years from now, instead of taking notes on paper, people would really have these tablet PCs. Making the PC easy to work with, dealing with speech recognition.

There's a lot of frontiers out there that we are invested in that we haven't really even begun to conquer yet. And so the explosion of the Internet, the interest in e-commerce, is almost creating a mania atmosphere in terms of funding new companies, and people wondering what they're going to do with it. You know, that is going to make all of our jobs pretty fascinating.

In the next 10 years, we can make a lot of progress on this stuff. We can make it far, far better than it is today. So, I don't see any lack of interesting things that I'm going to get to do in my job. You know, I get to come into work every day, work with smart people, we've managed to build a pretty incredible team. We have shared more of that success with our employees than any company in the history of business. It's like riding a bronco, we have to make sure we don't get tossed off. Success brings with it certain challenges, and we've got all of them in spades, including a lot of very intense competition.

So, you know, I think of this as sort of a hot seat, but one that I'm still enjoying quite a bit. So, you know, my life is devoted to working on software, that's the one thing I know and enjoy. So, all of my work will always be focused there.

SPEAKER: Okay. The gentleman in the back.

Q: If you were an ISV today focusing on knowledge management, and saw the directory structure starting to evolve, and the NT file system evolve, what type of products would you make?

MR. GATES: Well, knowledge management is a rich and complex area. It's almost a buzzword. The really exciting stuff is where you can go in and analyze the content of the documents, which is natural language type technology, and do automatic categorization for people. When you get email today, you're like a clerk deciding what folder to put it in, and which thing to pay attention, what to do with it. So, the email turns us into sort of low level actors, and that can take away a lot of time.

Knowledge management has got to let you step back and see those things based on the computer having some understanding of their content. Now, you know, that's been a holy grail for a long time. Nobody has really solved it. What people have done is basically text indexing in a fairly shallow way. And things like Altavista prove that that's actually pretty cool, even in the wide scale applications, but it's not really knowledge management.

So, I would either take a set of vertical industries, like the ones I mentioned, and just go and solve the heck out of the problems for those people, or I would take some deep technology piece, like the natural language, and go after that. Those are the two tacts that I think are fruitful.

Q: And how would you protect your margins as you go about it?

MR. GATES: Well, in software, margins are always super high, if I understand what you mean by the term "margin." The only problem we all have is our fixed cost of R&D. You know, our margins are all 80 or 90 percent, it's just you have this little equation, you have to multiply price times volume minus big R&D number, equals some positive number. And that's the balance to maintain. And so, you know, how to do you manage that? Well, there's three variables in that equation. You can either have the big R&D number not be super huge. You know, what's the magic there. Well, that's hard. You've got to have super-productive developers. Or you've got to have the volume be very high, that's the main number that Microsoft focuses on. We say, hey, if you can sell a million, you can figure out how to make money.

And then there's the price. And some of these solutions that people provide are of incredible value to customers, things like some of the data mining solutions, some of the configuration management software, things that really help them decide what loans to give, or how to work with customers, make decisions better, and so there are people who even at volumes in the hundreds or thousands, can charge a lot for their software. And that's great, you know, even though the hardware is going to be super cheap, there's going to be software that costs a million dollars for all time, because that R&D should be done. And so, you just have to decide what you want to make that equation look like.

For anybody in the software industry, there are some classic models of how much you spend in the different categories, but the fundamental decisions are price and volume, what's the sweet spot of the market, and therefore what kind of R&D can you afford. And, you know, in knowledge management the market is very big.

One of the problems we all have today, to be honest, is that capital is so available that the number of new startups in almost every category is gigantic. Go back three or four years, and how many people were starting up companies to do insurance software. Nobody. It wasn't some hot thing. Now, if you say insurance.com, you know, and you say, oh, it's an outsourced, Internet, XML-driven kind of thing, you can raise $100 million like that. You know, that's huge. But it does mean that in many of these categories, the heavy entry is going to create more competition in the next two or three years.

Now, capital won't always be available at the drop of a hat like this. Who knows when it won't, but there will come a day when it isn't. So we're all facing kind of hyper-competition in the next several years.

SPEAKER : The gentleman in the front.

Q: I would like to know what your strategy is on application hosting in relationship to true enterprise type applications versus the e-commerce stuff that you've been working on?

MR. GATES: Well, the NT platform is designed for hosting applications, and it's a key customer set, whether it's ISPs or ISVs, where we're saying, look, what do we have to do inside Windows NT to make it the best platform for you to do this hosting. Are we doing the right things with scale? Are we doing the right things with reliability? Are we doing the right things with the development tools? And so a lot of the applications will be hosted either on equipment that the ISP runs, or that some third party runs, and we've got to make sure our technology is the best at enabling that scenario. I think we're doing a super job of that. I think Windows 2000 is an amazing step forward. We have a lot of partners, many of them in this room, who are doing theirs using Windows NT as their foundation for hosting applications.

So we don't define it narrowly as just e-commerce. We define it as any application people want. We're going to give you a development approach where it works for Internet hosting, it works on corporate servers, and it works on portables. So, as you go to your customers and say, here's an application that does great things, if you want to run it as a service, that can be done; if you want to run it on your servers, that can be done; or if you want to let your salesperson in the field have some of this business logic, that works perfectly, too; if you want to mix and match, that all works. That is our proposition, because our platform is the only one that's designed to address all three tiers.

SPEAKER: To add to what Bill said, we're making major advances in SQL Server and Exchange to support that scenario, and we're also working very hard on changing our licensing policies, which is probably one question you were going to ask in terms of licensing both to you and to customers to enable that scenario to be done very easily.

MR. GATES: Yes. That's key. That's key.

Q: Bill, you're obviously the focal point of our industry and so much of what's going to be. We're all going to be hit with, at some point in time, which I'm hoping you can say roughly when you think that will be, when we can't really license software as copies, it'll be these browsers that just, you use it for a little while when you need it, and they don't want to download the software anymore, and it's just sort of, you know, a utility where, sure, we talked about it for 10 years, but it seems like with the network speeds and some day the browser appliance or the notepad, it may be a reality. When do you see that -- because for a strategic plan we have to think about changing our whole model, like you said, of volume times revenue. Do you see that happening in some time frame that we can anticipate?

MR. GATES: That will not happen. You're mixing two things together that I see a lot. The fact that the PC, where knowledge workers don't just sit in front of little forms that you design and say, I'm supposed to enter this number, and I'm supposed to enter this number, knowledge workers will not be sitting in front of a thing that is just a browser. Knowledge workers are there to think. All the jobs that are non-thinking jobs, where you just enter in data, those jobs are gone. This is an environment of total self-service. And, by entering forms into computers, you don't need that. Somebody calling up and saying, please, look up something on your computer screen, people don't need that. That will happen automatically.

So all the jobs that are left are knowledge worker jobs, where people need to see a rich set of information, and not just from one application. They don't just want to have this thing where, okay, now I'm running this application, let me just see that stuff. Oh, now I'm going to switch to this other application. They need to mix and match the data from those different things, to see patterns, make recommendations, to fulfill their role as a knowledge worker. So they're going to be creating documents, they're going to be editing documents, they're going to be annotating documents. They're going to have a device that works on their behalf, that has speech recognition, and a richer set of software functionality for them than any device that we've ever seen.

So the fact is that, yes, when they want to run an application it will look to them like they just click on the application and the application shows up. That doesn't mean that the code will all be running in some old time-sharing type approach, absolutely not, because that completely bans the idea that they can run when the network is not operational, that they can avoid overloading the network, that they can have local peripherals that work with that application, that they can take that machine with them and go and run the application when they're sitting down with the customer. And so the code, the code will often come down in a very big way, into that device, reduce the latency, give you the responsiveness, let you run offline. So knowledge workers are going to have a PC that is far, far more powerful than anything we've seen today. So that's the technical point.

In terms of the charging point, no, the people when a corporation wants to buy an ERP application, no matter what the technology is, they will pay just like they pay today. There's not going to be some new model where people say, oh, I don't want to pay for my ERP application, it doesn't work, nor are they going to say, I want to pay per purchase order I'm typing in. People don't want variable costs like that. I would never buy a piece of software like that. You know, Xerox a long time ago tried to sell copy machines like that, where you had to pay for every piece of paper that you ran through the thing. People don't want that.

Remember when everybody was going to go out and buy 50 components to make a word processor. They didn't really want a word processor that was tested together, documented together, they were going to buy little components and plug them all together, that was this big fad. Well, they also said that people were going to pay by the use, you know, I used the spell checker, that cost me a nickel, I deleted that character, this is nonsense. Yes, the price of word processors are today about a quarter of what they were five years ago. Well, that's competition. And that's a business I'm in, it's a very tough business. Fortunately the volume has gone up. But, there's nothing that breaks the basic equation of price times volume, minus big R&D costs. That's how people want to pay for software, when they buy the tool, they want to be able to use the tool at length, totally for everything they want to do with it, and not have to be charged on some ongoing basis.

So, code will run locally, and people will pay for software in largely the model they pay for it today. The only thing that's going to change is that instead of saying that you always have to host it on a server that you own, you'll be able to say to people, you can host it on a server that the ISV runs, or that some third party runs. That new model, where it's hosted there, will come into play. And people are already trying out all sorts of pricing models. But, with all these pricing models, at the end of the day you've got to measure how many users use your software, and multiply it by a number.

Whether or not code runs on their desktop or not, which it will, because that's the only way you get the responsiveness, that's the basic model. And so people like SAP or whoever, they think, how much can we charge per desktop for different classes of users, and you know, that will be their business model 10 years from now, 20 years from now, and that will be true for the vertical applications as well.

SPEAKER: Okay. Last question in the back.

Q: The question I have is really around this concept of the digital nervous system. It's an interesting concept. Where do you see the real time collaborative technologies playing in the digital nervous system, where do you see Microsoft going in the real time space in the longer term, and also what do you believe your position will be vis-a-vis Lotus?

MR. GATES: Well, real time collaboration will be a built-in feature of the platform. The fact is that seeing video and audio streams, getting rich echo cancellation, being able to share screens with each other, we've actually got that built in today, it's a feature called NetMeeting that when people discover it they think it's magical, that you can just use an Internet connection, and two people can work on a spreadsheet together, and depending on the connection you'll either do the voice thing by having a separate long-distance call, or if you've got a very good connection then you can do the voice across the same connection, as well. So all that audio streaming, video streaming type stuff, that's part of the Windows platform. Our primary competitor in that space is not Lotus, our primary competitor is a company called Real Networks, that does some of those same things.

So real time collaboration, at least the foundation for it, will be built in. Now, you have a lot of people who will come in and they'll say, okay, this lets you do meetings in a new way, and they'll innovate in terms of software that creates a user interface on what does a meeting look like. And so, for example, in the educational space, we have about 100 ISVs who take NetMeeting and what we've got there, and say, okay, what does a lecture look like, what does a discussion group look like, what does a lab look like, and they're all implementing a different vision of how education can be revolutionized by these things.

So there's not a big revenue for doing the plumbing piece, because although the plumbing piece is complicated, we've defined that as part of Windows. NetMeeting has been in there as a client that we shipped with Windows for three or four years now. And so it's the value-added applications where all the competition will take place.

SPEAKER: Okay. I think it's time for lunch. Thanks for the questions.

And thank you very much, Bill.

(Applause and end of presentation.)