Enterprise Solutions Conference
March 30, 1998

MR. GATES:
Well, good morning. I'm very excited about the group we've got here, and the opportunity to talk about where technology can be used to improve the efficiency of both business and government. We talked a lot about how the technology is advancing, but the bottom line for all of this is how it's used in an organization to allow that organization to do a better job. And our term for that is "digital nervous system." Every organization should move information in better ways by using the tools of the information age. In particular, the PC connected to the Internet.

Now, Microsoft was founded 23 years ago with the vision of what a low-cost computer could become as a tool. In the industrial age, we had tools that magnified our muscle power. In the information age, we have the PC, which, as it's connected to the worldwide network, magnifies our brain power. The advances in the PC have been phenomenal. Over 20 years ago, Moore's law predicted the doubling in microprocessor performance every two years. In fact, what we've achieved has been even better than that. So, the cost of computing has come down by a factor of a million during these last 20 years. And it's changed the nature of the computer. It's gone from simply a device that only large organizations could afford, to a device that every knowledge worker, and even every student over time, will be able to take advantage of.

Every aspect we look at, the microprocessor, the storage, the graphics, the screens, they're all advancing together because of the incredible volume and the open nature of this market. The term "open" is used a lot in the computer industry. But what it really means at the end of the day is customer choice. And until the PC came along, customers had a hard time because, once you picked a particular hardware approach, you didn't have any choice in the future because the software would be tied in to that particular hardware. With the PC, because of the volume, the choices in both software and hardware are the best they've ever been.

Some of these advances in the PC are easy to think about, the fact that we'll have so much storage that it goes beyond what you could type in an entire lifetime, and that extra storage will be used for things like photos, or even storing motion video. The extra speed of the computer will be used to make the interface far more natural than it is today. In the next 20 years, the price of computing will again drop by a factor of a million. And we're going to utilize that capability to make it easy to find information and to interact.

One of the most profound changes that will take place, that I think people underestimate, will occur as we get a higher quality portable flat screen device. That ability to take the PC and package all of it into a tablet-sized form factor that you can carry around with you, connected up to a digital network, I think will expand PC usage dramatically. Because you can take that PC into a meeting, you can use it to read information of all types, even the documents that you still read on paper today, and you'll be able to annotate those documents, send them to other people, use speech to interact with that device. And so, it will really be something that you're always working with.

One of the areas that's the greatest challenge in these technology advances is making sure that, as we connect to the Internet, we have low-cost, high-speed connections. Microsoft is very involved in working with communications companies of all kinds -- phone companies, cable companies, satellite companies -- to try and provide those connections. But that's the only area of the advances I have some concern that the technology will hold us back. Connecting up to businesses probably will not be a problem. There will be enough competition there that the organizations can be connected up. But when we think about connecting up to people's homes, or even to schools, I see, because of the regulatory issues, that that could move a lot slower than we'd like it to.

Well, let's turn back now to what the goal of all this technology is. Every organization has to deal with a lot of information. We call that the nervous system of the organization. Today, that information is largely on paper. There's a lot of phone calls and a lot of meetings. Simple questions like, "what's everything that's happened with this customer," are fairly difficult to answer because it's in many different locations, and there's not some powerful way of gathering that all together. Getting different parts of your organization that are in different locations to work together and share information is very difficult the way that things are handled today. I believe that by moving aggressively to put the information into digital form, and organize it according to the mission that you want to fulfill, the benefits can be quite dramatic.

Now, the information issues for a variety of things. You have your basic operations, keeping track of financial developments, being able to bill people, tracking your employees and what's going on with them. In these basic operations, a lot of it has been computerized in the sense that a central machine has a database. But it hasn't been made easily accessible to everyone in the organization. For example, let's say you want to see which of your employees have a particular interest or a particular expertise in a certain subject. Today, in most companies, you simply can't do that. Say you want to easily see what the latest organizational chart is, and what the background of the people involved in tackling a particular task is. That should be online. Whenever you're composing a mail message to a group, you should be able to browse that, and have a total awareness of that information. In most companies, that's very, very difficult.

Even things like printing out sales reports, my belief is that as soon as you take a number, something like a sales result, and put it on a piece of paper, it becomes far less valuable than when it's on the computer screen. Today, when I want to look at Microsoft's sales information, I simply call it up in a very summarized form, and with a few clicks, I can look at it by product, by country, by time period, I can do comparisons to what our projections were.

So, coming to this conference, it took me a few minutes to see country-by-country in Latin America how we're doing, so as I interact with our groups here, I'm totally up-to-date on what those trends look like. But if the information had been on paper, it would have been either far too detailed, showing me things that I don't care about, or far too summarized, where I would just look at the number, and if it surprised me, then I would have to call somebody up and ask about it.

As you can dive into the information in different ways, you can see the patterns, and then simply mail that rich view of the information that you have around to other people and ask them, you know, why don't we have a better result here, what should we do in a better way. And so, having sales information be easy to understand, cost information easy to understand, and having all the data about a customer in one place, I think that's a basic step to move toward a digital nervous system.

Another area of activity in an organization is planning events. And every organization has many examples of this. Things like the sales budgeting process, or the personnel review process, or project management. It's very important to have lots of information sharing during these processes, because many people are part of the activity. For example, take sales budgeting, we used to have a paper-based process that would talk almost two months to complete. And a lot of the time was wasted because you were trying to predict further in advance than you should have had to, and you didn't know what the other product groups or country groups were doing with their forecasts, and you always went back and changed things once you saw their work.

Today, by putting this online, we can compress the process into less than two weeks. So, it's about four times shorter. It takes away a lot less time. And, the quality of the information is much better, because you're constantly aware of other people, and how your information compares. We have certain checkpoints during the process, where we can see how things are going. So, it's a lot more fun process. We learn a lot more out of it, and have to put less into it.

Simple things like personnel reviews, getting everybody's input very easily, being able to check and do comparisons, by getting rid of the paperwork that's been greatly improved. Part of what the goal should be here is that when an organization is doing something similar to what it's done in the past, that you have all the work from the previous efforts easily available. So, if there was something that you did particularly well, or something you learned that you wanted to do in a different way, that should be accessible in digital form in a company-type intranet. So, that's an organization that constantly learns, constantly has the lessons and the previous work easily available.

A final area where having information in this digital form can make a big benefit is in unplanned events. The economy is full of surprises, thing are changing a lot faster than ever before. Things like currency rates, inflation, government policies, an activity by a competitor. These things come along at an increasing rate, and there's a real need to get people to pull all the information into one place, even if they're in different countries, even if they're in different organizations. And by having the information organized in the right way on an intranet-type approach, you'll deal a lot better with these things. I think every organization should measure itself by the ability, in many cases, to make a decision within 24 hours that's very responsive, and have it be a high quality decision that only these tools can really allow.

Well, what are the key technologies to do this? Well, in fact, they're the things that companies are already investing in. Most companies are buying state-of-the-art personal computers for their knowledge workers, and they're connecting them together into a network, and they have connections out to the Internet. So, the basic hardware is there. But the fact is that there's an incredible variation in how much value organizations are getting out of this. Every organization is using it to do word processing, or simple spreadsheet activity. But in most organizations, you'll find that the databases are spread around. People don't have an awareness of them. They're not connected together. And the key corporate processes are still driven entirely on paper.

One easy test here is the way an organization thinks of electronic mail. Lots of organizations have electronic mail, but they're not using it as a primary means of communication. And the reason is that until every knowledge workers is comfortable with electronic mail, and reading it several times a day, and that electronic mail system is being administered in a way that works incredibly reliably, until you get to that point, you don't have a useful electronic mail system. And so, people are always having to communicate in the old way.

Once you have extensive use of electronic mail, then you start to make very deep use of it, sending around not just simple messages, or short memos, but also sending around presentations or spreadsheets, and connecting that electronic mail up to customers and partners as well. This is the first step that a company has to take in terms of moving to a digital nervous system, to really incorporate electronic mail into their activities.

Another key step is to get the key information to be easy to find on the electronic system. One principle that I think both governments and businesses should adopt is that if you're going to publish a document, if you're going to put it on paper, you should also put it out on your intranet, because that's the place where it will be easiest to find. People can use powerful search techniques, and any of them with a PC will be able to see if that's a document that can help them.

In most countries, for example, the government is the biggest publisher of information. The government printing office is putting out a lot of brochures. And yet the people who might benefit from those, it's hard to know that it's been updated, and that it's out there for them, and there's just a lot of overhead to get access to it. Having that on the Internet can make a huge difference, and then you can even see which documents are being used. And you can even get feedback on those documents. Once something is on the network, having the ability to put someone's electronic mail address so that you can click and provide feedback, that becomes very, very simple to do.

One of the things we'll see in organizations is the elimination of paper forms. Already one type of paper document, the paper encyclopedia, is in decline, and the electronic equivalent is selling 10 times as well as the print encyclopedia ever did. The next type of paper document to go away will be these forms where you have to fill out lots of information. Inside Microsoft, I was fairly stunned when I asked someone to bring me all the paper forms. I found that we'd accumulated a lot of bureaucracy, about a thousand different paper forms. We had over 100 paper forms that related to personnel management, payroll options, adding positions, moving offices, buying business cards. And just understanding which form to use and filling out the same information, having somebody type it in later, that was all unnecessary.

And so, we eliminated every one of those forms that related to internal activities. We have no more paper purchase orders. If somebody wants to invoice the company, they go on electronically and fill out the invoice using an electronic entry system that we've put together. And so, we are moving to have data flow within the company without paper having to be involved.

Now, that doesn't mean very long documents, like the specification for a product. We're still printing those out. We're not able to go totally paperless, because we have to wait for that big breakthrough in the quality of the screen, and a very portable screen, before we're reading very long documents strictly without printing them out. But for the forms, the time is right now for everybody to move away from that.

As we get the high-speed connections, things like video conferencing will be commonplace, and electronic commerce will come into that. In fact, here at this event we're talking about some things we've done with Visa Latin America to create a framework for electronic commerce, and to make sure that in your countries that emerges very early on, and gives you access to customers on a global basis, and very efficient purchasing.

So, the vision I'm painting here is one of digital use of information, driving toward efficient governments and businesses. For example, filling out any of the forms for the government, that, too, should be done electronically. Governments, if they look, I'm sure they'll find they have a lot more than a thousand forms that you have to work with. And the redundancy that people deal with in filling those things in is quite amazing. Some countries have already moved to getting rid of the paperwork involved in customs. It's another great case where, to track things, search for patterns, make sure things are being done well, having it an intranet-type system is just dramatically better. Publishing information, as I discussed, it's very easy now to get things out in electronic form. In fact, we've made it so that when you create documents in Microsoft Word or the other tools, you can simply print it and publish it to the Web without having to do special work. So, it's one step, and you get both forms of the document that work quite well.

Now, it's very interesting that the United States, which has been the leader in so many aspects in using computer technology, in terms of its governmental use of computer technology and getting rid of paper forms, it's actually behind the rest of the world. In businesses, in universities, having PCs and use of the Internet, I'd say the U.S. is still significantly in the leadership role there. But, on the government side, there are lessons from all over the world where things could be a lot better.

Now, the first step for governments is moving information between government departments in an efficient way, starting with electronic mail. The second step is to make sure that the interactions with businesses, which overwhelmingly will have their own personal computers connected to the Internet, to make sure that those are being done without paperwork. The last step, and the one that's most exciting, is actually changing the way that you interact with citizens in allowing this electronic approach to be the dominant way things are done there.

One of the phrases that captures this opportunity is the idea of one-stop non-stop government, where you can easily do new things either by using your PC or going to a public kiosk. That's one of the big issues in terms of governmental use, is to make sure that every citizen can easily get to a kiosk to do these interactions. And so we see lots of examples where these are being put into the post offices or into the libraries or other government offices to make sure they are widely available.

The key point is that that kiosk is not just used for one thing. It's not just used for paying traffic fines, or taxes, it's not just used for medical payments, it's not just used for voting, it's used for everything where you're interacting with the government. And so you get complete sharing of the infrastructure, and the user interface.

Another key issue which is a major focus here in this conference is using this as a tool of education, allowing anywhere, anytime learning. By having access to the Internet, students can go out and explore information that they wouldn't be able to find any other way. Every student who is connected to the Internet has access to all that rich information. Also, there's a huge benefit for teachers. For the first time, they'll find it easy to share their best ideas, and put their material out online and have other teachers use that, improve it, comment on it and, therefore, benefit from each other's work in a way that simply hasn't been possible.

Even parents will connect in and see what the curriculum is, see what things are going on, and what they might want to talk with their children about. And so, tying in the schools into what we call a connected learning community is another huge opportunity where government and private companies need to make sure that great pilot projects are started even at this stage.

We will move eventually to also have universal identification, having some sort of smart card that allows us to identify who we are for these government interactions. Even things as sophisticated as electronic road pricing are already being done in places like Singapore, where you never have to stop at a toll booth, it's simply done with a wireless connection that they debit your electronic card, and so it's incredibly efficient. So, even communities will want to see information online and easily available.

I've got a number of examples here from around the world of where governments are taking a step to let citizens and businesses work in a new way using a digital nervous system. You'll see that I didn't pick any examples from the United States. The few that are there are actually at the state level, they're not at the national level. It seems that in many cases the smaller governments are able to get organized faster to be able to do these things. In fact, one of the smaller governments, Singapore, which I'll discuss, is the furthest along.

Two of the great examples in Latin America include the tax preparation filing that's being put together in Brazil, and they expect over two million tax filings this way this year, which is a pretty amazing number. And, the benefits to both the government and the people filing this way are very strong. In Costa Rica, they've been very aggressive in embracing technology, particularly in education. In 50 percent of the grade schools, they'll have PC labs, and 100 percent of the high schools.

Well, I mentioned that small governments are getting out in front in many cases. And Ireland is probably one of the best examples of this. Over half of their citizens on a weekly basis use the kiosks that they've placed in every post office. It's bill payment, banking, passports, pensions, everything that they can think of they've put onto this system. And their extra cost for adding new applications is extremely low. And so they're processing over two million transactions every day coming out of the thousand different locations where they're set up. Now, this will be done by every government. It's simply a matter of time before it gets pulled together.

Another example is down in Australia, not the Australian federal government, but a state level government in Victoria, which is the state that includes Melbourne. They've taken a lead in putting together kiosks that simplify your understanding of how to work with government. Instead of having to understand the different departments and the redundant information that they each want, they've organized it so you just simply say what the event is. So, if you're turning voting age, and you want to register, if you're changing your address, if you're getting married, if you're starting a new business, they guide you through the questions that they want you to answer, and then make sure that the appropriate parts are sent off to different government agencies.

In fact, when some of the information has to be sent to the federal government, what they actually do is print it out on the standard form, and then they simply send the paper in to the federal government, because the federal government isn't ready yet to actually have this information come in electronically. It's kind of strange because then, of course, that printed paper form is retyped into a computer later on, and probably in many cases that's a source of some of the inaccuracy.

So, here they're making government -- the face of government -- very straightforward, and they're having a very positive reaction to this type of capability.

Finally, the most advanced example is Singapore. Singapore, of course, has some real advantages in doing this. They've got a small physical area that's very easy to connect up with high-speed connections, they've got optic fiber now going into virtually all of the houses, and there's a requirement, as any new houses go in, that those optic fiber connections are actually provided. There's high use of PCs. We'll be introducing our WebTV concept into Singapore, and we expect to get over 90 percent of the houses using electronic mail and connecting up to the Internet to find information.

A lot of people have gone into Singapore to pioneer this. Some of their plans are pretty bold. For example, they'll take their entire judicial system and move it to be electronic. And so, you won't file papers in court anymore. If you want to file a lawsuit, track a lawsuit, you will do that strictly on an electronic basis. And, that will streamline their court system.

So, they're one government that can actually go to not only allowing paper and electronic to coexist, but also take the step of having the electronic means be the only way that the interaction is done.

Well, now let me move on to talk about some of the things Microsoft is doing to enable the digital nervous system. We view ourselves as having few very key products here. Windows, which is advancing in many ways. Our Office product for information creation. Our Visual Studio for developers. And our BackOffice to run the server and manage the rich information flows. Those products combined with our support and services are how we're creating a platform for the digital nervous system. And that platform has been used by software developers and companies to build the solutions that bring digital nervous system to them.

We have some exciting things coming up in the product area this next 12 months. We'll probably have as many major introductions as in any 12 month period. By the middle of the year, just a few months from now, we'll have Windows 98 out into the marketplace. About the same time, we'll have the final beta of the Windows-NT 5.0, which is the most important single product release that we're doing in this period. Also we, at that same time, we will have started the beta testing of the next major release of our Office product, and our SQL next release, which is Version 7, and Exchange. So, all of those pieces go together, and there's a lot we're doing with them to make them connect up for this digital nervous system.

In Office, I won't go into all the features here, but it's all about sharing information. People who've thought of office as just the way you create documents will be surprised how far we've gone to make it easy to find and share documents. We've also done a lot here on cost of ownership, making it so that you can simply do updates across the network without visiting machines. You can browse information up on databases with Excel. We see Excel as the viewing tool for all numeric information. And so, as you move from printing the numbers out to seeing them electronically, Excel will be the tool that helps with that.

Another key product for us is Exchange. And the big deal here is the way that we've integrated Exchange in with Office and with Windows. We took the directory that's in Exchange and we moved it into Windows-NT because people told us they didn't want to have multiple directories. We've connected up to all the other systems, including the ones based on Internet standards. The evolution that we're going through here actually has Exchange be part of the platform, and applications will be very dependent on it, particularly Office. Anybody who is using Office will certainly want to go down this path for building their mail systems.

We've very pleased with the progress we've had in the use of Exchange. Not only has it captured the majority of the market since it's been released just two years ago, it's also in extremely widespread deployment. We have many customers now with over 100,000 users working with Exchange.

I mentioned NT 5 as probably the centerpiece product. This is a product we've been hard at work on for several years now. And the key priority has been the whole set of issues around cost of ownership. How can we make it very simple to manage lots of PCs? How can we make it very predictable so that you know the PCs are going to work very, very reliably. How can we make it so that whenever there is a problem, you can just centrally see what's going on, and fix that problem without having to go to the user's PC?

So, NT 5 has a number of technologies that provide this. In particular, the rich directory and a feature we call Intellimirror. What people have been dealing with is that when you create information, a document, or any kind of information, your preferences of how you use the computer, historically all the information was stored on a very centralized computer, like the mainframe, and that had advantages in that it was easy to backup, and the professionals were there to manage that information. But, as the PC came along, having that information only in a central location really created big drawbacks, because it was very slow to connect up to the central information. You couldn't have applications easily get at the information, you couldn't have portable computers. And so, with the PC, we started to move the information more and more to be on people's desktops, stored in their PC. And, that, although it was incredibly responsive to end user needs, it meant that it was much more difficult to move the information around.

For example, if you would buy a new PC, how did you move the information, if your PC broke and you just wanted to plug a new one in, how did you have access to that information. And so, what really the market needed was something that combined the best capabilities of storing the information centrally, and the best capabilities of storing it on the desktop machine. And that's what we've done with the NT 5. We've made it so the information is stored in both places without the user having to think about it. To the user, it appears as it always has, although it's on the PC. But there's a replica of that information that's central, and so it means that if they do want to move over to someone else's PC and log in, they have complete access. If they want to upgrade their machine, they simply log in, and all that information comes down onto the machine. And so, this is a breakthrough in cost of ownership which we made the top priority for this work.

It's probably easiest to understand what's going one with our products if you think of the initiatives that span all of the products. In 1995, for example, Microsoft, after shipping Windows 95, we looked around and saw that the Internet was exploding in popularity, and a lot of people were saying that this was a big threat to Microsoft. They were saying that if we didn't get behind it and do a lot of work supporting the Internet that there were new companies coming along who would take over the role that we had had, and I think that was a very legitimate thing. Microsoft had to prove that it was very responsive, really going out, talking to customers on what they wanted out of the Internet, becoming the company that had more participation in driving Internet standards than any other, and really executing on the integrations very rapidly.

We're very pleased with how far we've come on that. Whether it's the server capability in Windows-NT, or the browser built-in, or supporting all those protocols, we have done a very comprehensive job. There's still work to be done. For example, working with Cisco, we're tackling tough problems like being able to send audio and video across the network, multimedia applications, without overloading the network, being able to track who's using the network, and making sure that the richest possible security is there. And so, by working with Cisco and then taking the results of those efforts into the Internet standards, we feel the Internet can become even better than it is today.

The other big initiatives for us are scalability, simplicity, and interoperability. Those are of critical importance, so let me focus in on each one of the three. Let me start with scalability. The goal here is quite ambitious, that is to make sure that Windows can run on the very simplest, lowest cost devices, and that it can run on servers that are powerful enough that even your most demanding applications can run there. And so, from top to bottom, through Windows, we have the same user interface, the ability to share applications development, and the same administrative interface to manage all of these systems. The range of capability is quite broad.

At the high-end, there are many exciting things taking place because not only do the microprocessor chips double in speed every two years, giving us that boost, but we're also increasing the number of processors that we have in a machine, going from a typical machine with two or four processors, to eight and 16, and eventually even to 32 or 64 processors.

We're also taking all of these machines and clustering them together. Today, this is just a two-node cluster where you have the fault tolerance, but in the future we'll go even beyond that for both scalability and very advanced reliability. So, Intel is driving things at the chip level very, very aggressively, and our key partners, like Compaq and HP are taking that and building really amazing server hardware for demanding applications.

Moving down to the other end of the Windows family, we've created some new options recently. We've created a thing called the Windows Terminal that allows you to run Windows applications where the actual application is running on the server, and so the desktop device is very, very inexpensive. For light users, for example, people just filling in information, that creates a new option that is less expensive, but fits into the broad Windows approach.

We also have some new smaller devices that are portable devices, or the WebTV that are also part of this same scalable family. These small devices are going to come in many different form factors. The handheld PC we came out with just about a year ago, and it's been very, very successful. I'm showing one of the new color devices that are actually in such demand that HP and others are having a hard time keeping up with the manufacturing that people are requesting for that. This uses Windows-CE, which is a sub-set of desktop Windows, and provides simple pocket type applications. We make it automatic to move the information back and forth, that's a big theme for us, so that your address book, your schedule, your messages are all easily accessible on this handheld device.

Now, there will be a new form of small device that comes out in a few months that we call Palm PC. Here we eliminate the keyboard to get -- so that we are small enough to fit into your pocket, and we use a touch sensitive screen with a pen so that you can select options, or even use handwriting to enter in text for this device. These devices will be quite inexpensive, under $500, and the screen quality that's available now is really quite amazing. So, they're a companion device. In fact, both of these connect up using infrared to send the information back and forth to your desktop machine. Eventually, the wireless connection will be a wide area wireless connection where you can go anywhere and the information will be kept totally up to date.

A very specialized form factor is the Auto PC. We showed our prototype of this at an electronics show at the first of the year, and we're amazed at the reaction we got. What we do here is, we actually build in speech recognition. The vocabulary is fairly small, so we're able to get quite high accuracy. And you can ask for traffic reports, or directions to a different location, or a radio station you want to listen to, or to have your cellular phone dial up somebody in your address book, all using speech. And by having the CD player as part of this, we can have all of the map type data accessible on that CD and kept very up-to-date. Now, this form factor will appear as a high-priced option to get with cars within the next year. But the big impact comes in about three years' time when we'll have the manufacturers building these in as factory equipment, even standard on high-end cars, and then on more and more cars as the price comes down.

Another new form factor is the WebTV. And this is where, instead of buying a special display, you use your TV screen, but you have a box with PC type electronics in it, that allows you to get out to the Internet. And we're working in partnership with phone companies in a number of countries, including a plan to make this available in Latin America as a low cost option. In fact, the hardware cost is very low, under $200. And so, the main thing that holds people back is simply the monthly fee. It doesn't provide full PC power, because you can't run local applications, but it does give you electronic mail, and connections to the Internet. And there's a lot of neat things you can do, like always have an up-to-date TV Guide, have more information about the TV shows, or follow-up information about the advertising, that's going to make this the standard way of watching television in the future. Television will be interactive, and it will connect up to the Internet, and make everyone's use of the Internet something that's just part of their everyday lifestyle.

I'm often asked, how far have we gotten in making the PC scale up to do the most demanding applications. And it's fair to say that any of the applications you want to tackle where you're willing to do new development, the PC now has the power to take it on. We've demonstrated over a billion transactions and there are many things that will drive those numbers up very rapidly. Our database performance today is 10 times what it was two years ago. Two years from now it will be 10 times, again. And so already, you know, keeping records for every citizen, and every transaction in the entire country, you can put together PC configurations that will do that very well.

We've also moved up to support very large databases. Databases have moved from megabytes, millions of characters, to gigabytes, which is thousands of millions of characters, and now in the world today there are perhaps 20 databases that are terabytes, in other words, millions of millions of bytes. And with the Internet, we're probably going to have a lot more of those very large databases.

Now, to support a database like that requires a lot of great support down in the backup capability, and the efficiency that things are done in. We've actually put together a fairly exciting example of this, that we call Teraserver. And I'd like to ask Chris Capossela , who is a Microsoft product manager, to come out and show us what this Teraserver is, and how it shows the scalability that we're building into NT.

MR. CAPOSSELA: Good morning, Bill.

Good morning, everybody.

(Applause)

MR. CAPOSSELA: Thank you.

Microsoft Teraserver is a pretty exciting project. We've basically asked ourselves, how big a database can we run on a single PC. So Teraserver is a 1.4 terabyte database, running on a single NT machine. That 1.4 terabytes of data is stored in a single SQL Server database. Now, a terabyte of data is a really hard thing to get your arms around, because it's so huge. If you took all the Web pages that are out on the Internet today, and copied those down to your hard drive, that still wouldn't be a terabyte of data. If you take the New York Stock Exchange, and take all their transactions that they've ever done, that would still only be about half a terabyte of data. So there are very few databases out there that are this big. And in trying to push the envelope we had a hard time actually figuring out what kind of data we can use to build a massive database.

We wanted to pick something that everyone can relate to, because we plan to put this up on the Internet in June, so everyone around the world can play with it. So we decided to build the world's largest atlas, so people can go in and see different regions of the world. And we partnered with two interesting organizations, the United States Geological Survey, and the Russian Space Agency. And these two organizations just have incredible satellite and aerial photographs and images of much of the urban areas around the world. The U.S. Geological Survey actually has great U.S. data, and kind of interestingly, the Russian Space Agency has some pretty great data on the U.S., as well. But, they've got some great data on other regions of the world, like, Hong Kong, and Athens, and Rome.

So once we had all this data, we got about 5 terabytes of raw data from these two organizations, and we tried to take just the cream of the crop and put it into our SQL Server database. But, once we had this massive amount of data, we needed a way for people to be able to find the data. So we took the index out of Microsoft Encarta World Atlas, and that index has about 1.1 million records in it, it knows about 1.1 million places on the planet. And we put that in the database, as well. And just to give you an idea, those 1.1 million records take up about a gigabyte of storage, as opposed to the terabyte, or the 1.4 terabyte that the images take up.

Now, once we have those two together, we can create a little Web UI here, that you're looking at on both screens. So you see, we've got a map of the world, so anybody can just point and click to a region of the world that they want to see, and zoom in, and see the area that they want. We also have this thing that we call Picks of the Week, where we basically just highlight a few interesting images. Since we're from Seattle, I'll go ahead and click on the SeaTac Airport here. And for any of you who have ever flown into SeaTac, this is a pretty amazing image. You'll notice it's from the U.S. Geological Survey. And you see, we're looking right at a runway, here at SeaTac Airport. And we can scroll around, and you notice the interface actually let's us pan up and down. So we can kind of click our way across the country, if we have plenty of time.

The amazing thing about this is, this is one meter resolution, which means every pixel on this screen takes up -- or is relative to one square meter of space. So, for instance, I get incredible detail on this plane, I can see that that's a 747, because I can see the two engines on each one of these wings. And I can see another plane taxiing right here, to take off. If I scroll up, I can click the zoom out button, and now again, every time we request an image, think about what's happening. Our machine here, which is just running Internet Explorer 4.0 and is connected to the Internet, it's sending a request out to the Web server that's running on our Teraserver. Again, it's just one single machine. It sends the request out to the Web server, and the Web server then queries the SQL Server database. The SQL Server database builds an HTML page on the fly, sends that back to the Web server, the Web server sends that back over the Internet, and it's displayed for us right here.

So now we have a little bit of a higher level view, and I can see they are actually two runways at SeaTac. And even from here, I can sort of see a little image there, let's click on that and zoom back in, and I think we'll see a different plane in the process of taking off. Now, we've taken these big images, and we've actually sliced them up into little frames. Yes, there it is. It's about to take off, it's looks like an MD-80, because you can see the two engines at the back. Being from Seattle, where Boeing is located, I get to sort of know the planes a little bit better than your average person.

We actually split these up into little frames. If I go ahead and click on, let's say, the search button, actually, let's go up here, and go to Teraserver search, this will give me another area of the world that I can click on. And you can see the green dots represent places where we have images. But, if I scroll down, I can actually type in the name of someplace. So I'm going to type in the North End, which is where I'm from in Boston, and type in USA, and hit search. And again, it's going to go off and bring back the various North Ends that it knows about in the USA and show us the images that it has from the Geological Survey, or from the Russian Space Agency.

Let's click on that one. Now, we've split these images up into little tiles, so that we can download each one of these tiles a little bit faster than it would take to download the entire image. Those tiles are also nice for us, because it means when you scroll up and down -- so there's Boston, right here, pretty amazing. Logan Airport is over here. The North End is right there, in the clump of all these buildings. I can zoom in a little bit. But, those tiles, when we scroll around, left and right, they make things a little bit faster, because we don't have to get all the tiles again, we can keep half the tiles, and go get the next half, the ones off to the left, or off to the right, or up and down. So, again, I can make this image a little smaller, a little bigger, if I want.

The final thing I'd like to show you is, let's go to Disney World. Here we are in Florida. This is actually an image from the Russian Space Agency, where all the others have actually been from the U.S. Geological Survey. And we can take a quick look at Disney World. Now, these two organizations were very -- they're great partners to work with, but they didn't just want to put this information up on the Web, they also wanted to make it easy for people to actually order the image. So we can come in here and say we want a small image, or a medium image, or a large image. And there's the price.

These organizations have used Microsoft Site Server to actually allow this transaction to happen. Now, I can say I want this in JPG, or in TIFF, the different file formats I might want, click on next. And now the Russian Space Agency has even allowed us to -- or even set this up so that the user can actually print things and have that information sent to Kodak. Kodak will take the image, actually give you a print out of that and send it to you.

In this case we're just going to go with the image that we download over the Internet. Again, we're using Site Server, Microsoft Site Sever to do the credit card authorization, and make sure that all of that goes very, very smoothly. Telling me how much it's going to cost, asks for my credit card information, and we can go ahead and purchase it. So it's a nice example of some great scalability, but also sort of some e-commerce here, and these folks obviously have incredibly low distribution costs, and they can do just massive amounts of volume over the Internet.

Now, I should mention the hardware that this demo is running on, we're actually running on a Digital Alpha 8400 series. Let's go back to our home page. And that 8400 system is pretty amazing. It's got 8 processors, each one of those processors are 440 megahertz processors. And the machine itself has 10 gigabytes of RAM. So this thing is a very large machine. But, again, it's a single machine, running on NT 4.0.

Now, you see my desktop, I actually have a picture of an early version of the Teraserver, back in Redmond. This Digital 8400 series actually has 8 cabinets, just like these that house it. One of the cabinets is for the eight processors, and the other seven cabinets are for all the hard disk space. We have 324 hard disks across these cabinets, and they give us about 3-1/2 terabytes of raw data storage. So it's a great example of scalability with SQL Server. Actually, the next version of SQL Server, we're kind of beta testing it, and it's been a great beta test. And it's running on NT 4, and off the shelf hardware. It's really changed the economics of building a database of this size.

MR. GATES: It's almost like a mainframe.

MR. CAPOSSELA: It looks that way, but very simple to administer and much cheaper.

MR. GATES: Great. Thanks a lot, Chris.

MR. CAPOSSELA: My pleasure.

(Applause)

MR. GATES: The use of NT for mission critical systems has gone up dramatically in the last two years. When we talk about NT, of course, we include BackOffice. That's our term for the product that combines SQL, Exchange, and our system management. I've got some great examples of customers doing some leadership things, using this combination. Bank of Credit, in Peru, has connected up both the branch and call center solutions in a solution that's been a big money saver for them. Disco SA, in Argentina is using it in their supermarket chain for point of sales, managing inventory in a better way. And once they had things up electronically, they decided it would be easy to go out on the Internet, and so they're one of the first to be out there.

In terms of banking, banking in Latin America, in many cases, is more sophisticated than in any other part of the world, because there has always been a need for responsiveness, real-time type banking. So, in fact, Internet banking is as advanced, particularly in Brazil, as it is anywhere in the world. Prudesco (sp) has been very strong on this. And they've already got over 200,000 customers. That's more than any bank in the United States has hooked up. In Argentina, Maxima (sp) AFJP, provides access to pension fund accounts. So somebody can come up, see what's going on, make a change in what's going on. It's a very large database, and again, there's money savings associated with this.

Today, in the world of computer technology, projects are not as long as they used to be. Historically, you used to have two and three-year projects, and what you were interested in may have been out of date by the time it got done. And some of those projects simply never got done. Today the projects are far shorter, because the tools are better, the building blocks are much stronger, much richer. And many of these projects can have a goal of both doing things more efficiently and being less expensive than what came before.

Now, I mentioned another big initiative for us, beyond scalability is simplicity. And, in fact, I'd say today this is probably our top priority, because we've done a lot with PCs that make it so you have to learn way too many commands. Things like upgrading the software, learning the software, finding information. The commands you give to search are very different. If you want to search your mail, it's one thing, if you want to search the Web, it's another thing, you want to search for files it's another thing. And many of the messages are very hard to understand.

For example, this is a sample message, that when we were reviewing all of our error messages, we thought, do even we understand what we do when we get this message? It says, the DHTP client could not obtain an IP address. Well, users shouldn't even know what those words are. It's so helpful, it says, if you want to see the DHTP messages in the future, choose yes, otherwise, choose no. Well, that's really helpful. You know, what does it mean. What if you say no? What if you say yes? And we simply can't have things like this. We've got to automatically diagnose any issue. You know, this box should say, we couldn't detect a network connection, did you expect that this machine was connected up to a network? If you say yes, then we start to diagnose why the connection isn't working, and work with the user, so that error messages and the help system become integrated in a way that it's always directed at helping the user out.

Unification of different parts of the system, moving to one interface, is going to be a big way that we improve things here. With the new releases of Windows we're doing this year, for example, we've eliminated our help system. We used to have a special system called Winhelp, that had its own commands, its own file format, its own utilities. It's gone. We're simply using the browser. Now, we had to add some features to the browser, to make it capable of taking over there, but they turned out to be things that were very useful for all information that you want to browse. And so we are making steps towards this simplification. The next big step is to merge in the mail, and the file system, so you browse those in the same way. And we're already hard at work on doing that.

I did want to hit on one competitive thrust, which was called the NC, or Network Computer. This was a big thing when companies started to recognize the high cost of ownership of their PCs. Some of the companies who haven't really embraced the PC movement, well, this is their opportunity to come along with something different. And so this NC concept was promoted, particularly by Sun, Oracle, and IBM. Now, the whole cost of ownership focus, we agree that was a good thing. In fact, it was very important to start to see how people were spending money, and be able to reduce it substantially. But, it turned out the NC really was a dead-end here, because it required people to rewrite all their applications. It required people to learn new interfaces. It required you to run almost everything up on the server. So you gave up the richness, the portability, the responsiveness of the PC.

The claim that the NC would be less expensive didn't turn out to be correct, because the PC, as a very high volume device, is always going to be the lowest cost way of running applications. As long as you have something like the browser, which is the biggest application running on a device, it's going to be as expensive, or more expensive, than a PC. And the NC turned out to be that way.

Fortunately, the press has noticed this, although at first they were kind of excited about the idea of something new. There's been a pretty strong recognition here that that hasn't happened. But, in the space of the PC, we have made some good progress. On the hardware side, hardware prices have come down faster than ever before. And we've come out with administration kits that help you manage things, so you can just control what options a user is working with.

The Net PC is an approach that Compaq and HP have been great partners in working on the design of that with us. The idea there is that it's a box that you don't have to describe what the hardware is. It's automatically detected, and you don't have to open the box up, and change things. And so that simplifies the hardware side.

The big advance will be NT 5. Everything we've done up until now is not nearly as impactful as that is. But, even so, if you look at the latest Gartner numbers, they now show a well-managed PC as having substantially lower ownership costs than it would have had in the past.

Now, if we look at the computer industry, it has really been changed by the arrival of the PC. I've got here pretty much the top companies, in terms of size. One of the things that always surprises people is relative to these other computer companies, how small Microsoft is. Microsoft has incredible visibility, because of the role of our software. We took a very low-price, high-volume approach to software. And so we've been able to set some central standards that have impacted the entire PC activity.

The interesting thing here is if you look at all these leading companies, you can say, over the last several years what has happened in terms of their involvement with Windows NT. And every company up here, with two exceptions, have moved their strategy to embrace Windows NT as a very central part of what they're doing. These companies see Windows NT as the high-volume platform, both on their services, and in their products, are doing a really great job of embracing that. Here in the United States, the two leading partners are certainly Hewlett-Packard, and Compaq, who have been doing great work along both NT services and products.

I said that there were two exceptions on here. Sun is a pure exception. They do nothing with Windows NT. And IBM is sort of a mixed situation. They do quite a bit in PC hardware, but their strategy still doesn't have the PC technology at the center of activity. So some groups there are involved with it, some groups there are still pushing into different ways. But, the industry as a whole has taken on a dramatically increased NT type focus.

Partnerships have been key to what we're doing. I've just listed here some of the key partners, in no particular order. Because we are specialized in developing just platform software, we need partnerships to make sure that our building blocks come together into solutions. Even in the area of software, we've chosen not to do enterprise high-end software. And that's made it very easy for us to work with SAP or Bonn (sp) or PeopleSoft, and get them to come in and build on our platform, in a very rich way.

In fact, if we just look at SAP, they've gone from having no installations using PC technology just a few years ago, to now the majority of their new installations are built on top of Windows NT. And so that's been a real showcase. In fact, Microsoft itself is using NT and SAP to drive all of the things that we're doing. I mentioned earlier, Visa Latin America, which is a key partner, in terms of electronic commerce, and you'll be hearing more about some of the exciting work we're dong with them.

A final initiative I mentioned is interoperability. Despite the fact that a higher and higher percentage of systems are built around our platform, we recognize this is a world with many different types of systems. There are going to be mainframes, and high-end UNIX systems out there basically forever. And we're going to have to make sure that we can exchange and interoperate in a rich way with those systems. There's been a lot of neat developments in this area, for example, our COM object model is now available on UNIX and Microsoft itself is doing the development and the support of that. We're connecting up with Cisco, to make sure the networks can tie together using our Active Directory, and making sure there can be simple management there. We've also done a lot of things to reach out to the databases that you're going to have on these machines.

Let's say, for example, that a bank has decided that it's going to leave its master database on the mainframe. But, it doesn't want to write anymore mainframe code, because the tools just aren't as good, you don't have the Internet type approaches. Well, now, because of the gateways we have for going to get the data out of IMS or DB2 databases, you can do that. In fact, we're finding that a lot of banks now have made that distinction of leave the data there for the time being, but move all the new applications onto the NT platform. One big milestone here was having our transaction server connect into CICS in a very rich way. So interoperability just gets stronger and stronger, because Windows becomes a hub that is able to connect into all these different systems.

The products alone are just one part of the picture. We've got to have the support and services around that. Even though we ourselves don't have a goal of making money in the services side, customers have really asked us to step up to 24-hour support and having more people who can help review the architecture. Now, the primary strategy we have here is to work through partners, to get them trained and certified. And so the fact that we're approaching 200,000 certified professionals, that's the thing that is most important here.

Now, there is a shortage of skills out there. I was talking with some customers last night, and that came up as a key topic. How do they go out and find people who have got the right background to help them build these new systems. We're certainly going to have to do a lot, working with the universities, working with training organizations, to make sure that there are more people who understand NT and building Internet applications on top of it. And we've got some creative ideas about how we can do that in Latin America.

This is an area of the world where we made an early commitment. We maintained that commitment through all the economic expansions and contractions, because we take a very long-term approach to the things we do. And so building those partnerships year after year, helping the local software developers, with the latest information, which is a thing that we've expanded quite a bit this year, and then working with the companies in the country of all types, that's worked very well for us. We are seeing very rapid growth and the only thing holding us back is that skills gap, and so we'll certainly address that as a very, very high priority.

Let me just finish by talking about some of the opportunities, looking forward. There is an incredible amount that can be done to make the PC better. We're constantly increasing our R&D. You can see our R&D percentage is moving up toward 20 percent, which I think we'll get to in the next year. And that is certainly one of the highest. There are a lot of breakthrough technologies, speech recognition, handwriting recognition, visual recognition, ways that the computer can help out. 3D graphics will become part of the interface. And we haven't done that yet. The computer will be able to learn, in terms of what you're interested in, and help find things for you.

I'd like to, just real quickly, show you an example of how intelligent assistance is going to come together. I'd like to ask Chris to come back out again. This is a demonstration that actually combines a lot of different things, some natural language recognition, as well as with some speech technology.

MR. CAPOSSELA: All right. I'm back. I've been told by the translators that I have to go much, much slower. So I'll try.

Basically, what we've done here is to take Microsoft Outlook '98, which is the latest version of Microsoft Outlook, and combine it with our Microsoft agent technology, along with text to speech, and speech recognition technologies that our research group is working on, to build this prototype. So today Office has an assistant that's a good step in the right direction, but it doesn't learn well over time. It knows how to do certain things, and it doesn't really learn the way I like to work.

So with this new assistant, or this prototype assistant, I should say, we've basically had it watch me open about 800 different emails, and see how I do things with those 800 emails -- what things do I schedule, what things do I just read and reply to -- and kind of offer me help at the right time.

So here is an email from Eric, who says, hey I'd love to meet with you, but I've got a bunch of things to do this week to get ready for something we're doing next week. Can I take a rain check on the meeting, and try to schedule for a week or two later? Now, when I open this message up, our agent is going to parse this text. And it's going to see information.

So the agent comes up, does its thing, and goes away. You can see he's obviously talking to me. He's listening for my answers. And in this case, since the person asked me for something a week or two in advance, it brought up the calendar and it highlighted the right week for us, and it brought up the calendar in the weekly view. Let's go ahead and close that down, and let's close this down, and take a look at another message. The next one says, hey, how about arranging to have a big review sometime down the road? How does your schedule look in June? So open that up, again, it will parse it. So again, it gives us a different view this time.

And finally, the last example I have is a much more concrete one. In this case, Eric says, hey, how about if we do a conference call Saturday at 4:30 Seattle time? So the agent has gone ahead, created an appointment for me, put the text of the email in it, included Eric on it, and I can just go ahead and hit save and close. So it looks very natural here, but if you think about the steps that I do today to make that happen, this is replacing like 15 or 20 different clicks, where I have to go off and check Eric's schedule, manually, to see when he's open. It's really making my life much easier.

Now, we haven't really illustrated how the agent learns over time. But, again, it's sort of watching as I work with my incoming email messages, and if it sees a very long email message, then it's going to wait and let me read that for a while and see how long it takes me before I take my action. And it uses all that information to judge when to help me out.

MR. GATES: I think intelligent assistance is going to be very important. Thanks, Chris.

MR. CAPOSSELA: My pleasure.

(Applause)

MR. GATES: Let me just close by talking about some of the opportunities. Clearly, we're very excited about the opportunities to have better software fit in, and let you build digital nervous systems. We're very focused on enterprise requirements, large organizations that are dealing with tough problems. Partnerships are very central to this, and today the event we're having here includes most of our very key partners. So we think now is the time to invest in digital nervous system. And we're anxious to have your feedback, and very much looking forward to working with all of you.

Thank you.

(Applause.)

MODERATOR: Thank you, Bill. Bill, there is no doubt that the implications of this kind of technology covers the entire society and actually, I would like to acknowledge the presence of Mr. Ernesto Diperez Bialaris (sp), the President of Panama, who is today with us.

Thank you very much.

Okay. It's time for questions. There are three mikes, in the corridors. You can make the question in Portuguese, Spanish, English. Bill has a device. So I'm going to start with the first one, that I just got over there.

Bill, are Java Beans a threat to Microsoft?

MR. GATES: Okay. There are a lot of different technologies for connecting objects together. And there would be probably two or three that are used enough that we need to make sure that we have gateways that connect those. CORBA is out there, some people are using that. COM, which is the one that we built into Windows, is by far the most popular, and we're doing quite a bit to evolve that. We did make a big advance in letting COM and CORBA systems work together, when Iona, who is the biggest provider of CORBA technology, agreed to highlight COM and include COM interoperability in all the products that they make available.

Java Beans is another approach that's not compatible with CORBA, it's just a different approach. And we're also building gateways that will connect up to that. So whatever the important standards are, we're going to connect to. Java, in particular, is a language we support. Now, we're not trying to be religious about this. We know that next year there will be some new computer languages, we'll support those as well. Visual Basic and C will continue to be by far the most popular two languages, so all of our work is language agnostic, including having a great Java product, that is the most popular one that people use today. So any different language and COM is our strategic approach to object connection, but we're connecting up to all the other approaches as well.

MODERATOR: Okay. Thank you.

I invite the audience, if any question, the mikes are open.

I have another one. Is Microsoft -- will Microsoft be an Internet service provider?

MR. GATES: The answer is no, we won't. We are very anxious to be in partnership with everyone who will be an Internet service provider. So we're not going to own the fibers, or the cable, or the satellites, any of those things. We're sitting down right now, though, to say that anybody who is doing that, we want to help them set up a hosting business, we want to help them with the browser, and we want to really drive the market, whether it's improving the PC, or coming along with things like the Web TV. And so we've actually got a customer unit that is dedicated to working with those network operators.

MODERATOR: Okay. We have the first gentleman over here, please.

QUESTION: Large- and medium-sized banks in Latin America base their mission critical systems in IBM mainframes and IBM operating systems and databases. You mentioned something in your conference, but can you expand on how you foresee the evolution of the information systems of these banks in the next few years?

MR. GATES: The banks have been great customers of Microsoft, because of course, banking is an information business. And moving information effectively, deciding which loans to give, doing that quickly, that's central to their competitiveness. And so they're broadly using PC technology. In fact, if you look at the branch banking in Latin America today, all the new people who have rolled out things in the last years have been using Windows NT. Basically, OS2 has completely disappeared from that role.

If you get to the very central machine, the mainframe, there are two approaches. One approach is simply to replace it with PC technology. And there are medium-sized banks that have gone ahead and done that. Most banks are going to be quite conservative and wait a number of years before they take that step. And so what we have to do is show them how all the new code they write can be on NT, and yet have it interact in a fully transacted way, with that mainframe data, which is actually in most cases not even DB2, most of it's IMS-type data.

And so whichever of those it is, that was the central goal of our transaction server. And so we really want to work with banks to see how we can move their development down onto the NT, where you're going to have the best Intranet tools, and be able to pull the information into very rich views, both for internal use, and for your customers, that over time will be connecting to you through the Internet.

MODERATOR: I would like to add that today in Latin America, 35 percent of the branches, banking branches are using NT, so far.

Okay. We've got another question over here, please, gentleman?

QUESTION: In 1996 Andy Grove extended the -- (inaudible) -- life of the semiconductor into the next millennium. Do you have, Bill Gates -- (inaudible) -- for the first 10 years of the next millennium?

MODERATOR: Can you repeat the question?

QUESTION: The question is, in 1996 Andy Grove extended the -- (inaudible) -- life of the semiconductor into the next millennium. Do you have, Bill Gates -- (inaudible) -- life for the next 10 years in the next millennium?

MR. GATES: Okay. Well, the -- Andy Grove has made it clear that Intel doesn't see any slowdown in their ability to build very high-speed chips. And, in fact, Intel is often coming to Microsoft and saying, what are you going to do with all this speed we're going to provide? Can your software take advantage of it in a nice way? And so we've spent a lot of time showing Intel our research work on handwriting, speech recognition, visual recognition and learning technologies, which in fact require those new processors.

The prediction I would make, which is very aggressive, is that within the next decade most business will be done on the Internet, that people will have these tablet-sized devices that will be very natural to use. And even in a meeting like this, most people will come with their tablet-sized device, and be able to take notes, and look at information, and find that a much easier way of doing things. So that's where I think people are underestimating the Internet, and they're underestimating, as we get to these new form factors and natural interfaces, how widespread all of this will become.

MODERATOR: Thank you, Bill.

Another question over here, please?

QUESTION: Hello. My name is Claude Orago (sp) from the Chilean government. Mr. Gates, despite the amazing developments of computer technology, there are still many concerns in both the user side and the industry side about the future of information technology. From the users there's concern about security, supply driven demand and things alike. And from the industry, there are concerns about regulation, lack of political will from different governments. So my question will be, what are the main risks and obstacles for the development of information technology in Latin America, and in the world in the future?

MR. GATES: I think there are a number of things that we need to be concerned about. The first is the skills gap that I talked about. And every country has to look at, are its schools and universities turning out people who are familiar and comfortable with these tools? Most of the great jobs that will be created in the future will involve using the computer in some way, not just in the information technology industry, but if you look at an industry like car making, the things that count, in terms of great design, great planning, they will need these tools. So I'd say the skills gap would be a big issue.

I'd say having a high-speed communications infrastructure, that reaches out to all the citizens, that would be a huge problem that each country will have to deal with in their own way. I think the issues of privacy, we can come up with very good solutions there. The technology does allow us to make the information secure. There are some political problems where the United States government is not yet allowing us to export the best encryption technologies that can be used for privacy. And so Microsoft and many of the other companies here are pushing for some liberalization there. And if that doesn't come we'll have to come up with alternate approaches. So privacy, a lot of people will be very concerned about that moving forward.

The final concern I'd mention is that, given that this tool starts to fulfill all the exciting things I'm talking about, then it becomes almost as important as literacy, where you actually want every student to have access. And that's why we're really pleased to see that there are some great pilot projects, in the region of getting more PCs out into the schools, even for younger people. And so both schools and libraries will be a key place, and it will require more resources to make that happen.

MODERATOR: : Actually we have these pilots in Argentina, Brazil, Costa Rica, Mexico, and Colombia, as well.

Okay. We have another question here, please.

QUESTION: Yes, Mr. Gates. You spoke a moment earlier about the skills gap. I want to talk to you a little bit about the Microsoft certification program. Those that go into the program now can expect to spend anywhere between $6,000 and $10,000 to get a Microsoft certified trainer to train them. Then, they can take the courses of which perhaps only 70 percent of what you'll find in the test is actually included in the books.

Now, I understand you have said at conferences like this before, that you intend to keep the MCSE a step above the CNE, and to make it a certification that only the highest qualified professionals can obtain. On the other hand, you're talking about trying to fill this skills gap to get as many qualified people out there. These two things seem a little contradictory. Could you tell me a little bit about your plans for Microsoft certification in the future?

MR. GATES: Okay. Microsoft has a number of certification programs. The most challenging, the highest level, is what we call the MCSE, Microsoft Certified System Engineer. We have made that one very tough for people to get, because if you want to hire one person and say, manage my servers, make sure they're secure, make sure they're up 24 hours a day, that becomes very critical to your company. You've got your electronic mail running through those. You've got your electronic transactions running through those. And so, we don't want to have any confusion about who's really got the right background to manage these systems, and to manage development projects around them. So, we've kept MCSE very high.

Now, there are other levels of certification. You can be certified for Visual Basic, you can be certified for Office, or Exchange. And, so if people want to be narrowly focused, we have the other certificates. I do think there is a lot we can do to bring these training costs down. In the United States, we've gone into partnership with a number of the universities, particularly what are called community colleges here, and created much lower-cost training alternatives. And I think that's certainly got to be done in Latin America. We'll try to be innovative on that, work with our ATECHs (sp). We have opened up some of the course material authoring now. And that, I think, will allow the prices to come down somewhat.

But, in the final analysis, there's no escaping the fact that the person who manages these systems and picks the tools to do development, you want some designation that gives you a lot of comfort that person has a very in-depth background. And so, we are dealing with a challenge to which there's no simple solution.

MODERATOR: : Okay. We'll have the three last questions, please. The first one is over here, please.

QUESTION: Okay. In Spanish, please.

MODERATOR: : Okay, he's going to make it in Spanish.

MR. GATES: Great.

QUESTION: (In Spanish.)

MR. GATES: Having rich security, there's two big problems there. The first is to make sure that whoever walks up to the system to start the transaction, that you really know who it is. That's the identification problem. And there, today we use passwords. If you want to get into your electronic mail, you give a password and you have access. In the future, that will move to a combination of having a smart card and having a password both. So, that increases the security. It's a lot like is done with cash machines. There are some alternate proposals, where you would use fingerprints, or voice prints, but it's too early to say that those technologies will be in widespread use. So, I think the smart card plus the password is the clear direction there.

The second problem in security is making sure that no one is tapping in, so that no one is listening to that conversation, whether it's a video conference or an electronic transaction. That's where we use these rich encryption technologies --

(End of tape 1.)

MR. GATES: (In progress) -- problem. And so, simply, if you administer the system well, you shouldn't have to worry about people tapping into either of those kinds of transactions.

MODERATOR: : Thank you, Bill. We have time for two more questions. The first one here, please, and the second will be here.

QUESTION: Okay. Marisa Scandona (sp), Venezuela. My question is related to Internet telephony. Older application for Internet telephony requires some special software and were from computer-to-computer. Now, with IP applications, we can have phone-to-phone and quality to spare compared to the older application. How do you think this is going to impact IP applications, and IP telephony and special telecommunications companies, and how is Microsoft supporting Internet applications, IP applications?

MR. GATES: Great. IP telephony is going to be very popular. And we include in with Windows a product called NetMeeting that lets you connect up, not only voice connections, but also share your computer screens. Two people can be working on a spreadsheet or a document at the same time, or send a video channel. The tough part is making sure that the connection between you and the person you want to connect to, that you have very high-speed connections, and that other users aren't coming along and using up that capacity.

That's why phone calls today on the Internet, if you just use the general Internet, not a special connection, the quality is very low, because if a lot of other packets come along, then it doesn't get delivered on time. That challenge of having a standard way of being able to use the general Internet, that's one of the key priorities we're tackling in partnerships, particularly the one we have with Cisco. And, so although today you have to set up special connections to do IP telephony with high quality, in the future, we'll make it so you can do that over the general Internet. There will be some extra charges to be able to prioritize your packets, and so there will be a new tariff structure that goes with that.

This is one of the great challenges that governments have to face. As voice calls move on to the Internet, how does their tariff structure adjust to that, because today they're taxing the voice calls, and in the future most of those will be done across the Internet. Even in the United States that's a big issue that hasn't been fully worked out, but there's no doubt that IP telephony will be the main way people connect.

MODERATOR: : Thank you, Bill.

The last question, please.

QUESTION: (In Spanish)

MR. GATES: Great. Well, we do have some special sessions as part of this event that are going to focus in on education. And, you're absolutely right that this is a very tough problem. There are no simple answers. You know, the ideal would be if we could give every student a laptop computer, because then they own the computer, they can take it home, use it as much as they want, and this has been done in about 400 schools now, and the results are quite impressive. But it's out of reach, even in the United States there's no way that can be done broadly in the foreseeable future. What needs to be done is to have computers that you share in schools and libraries. And it's not just having the computers, you have to have the connections, you have to have the teachers enthusiastic about that, and you have to be able to share best practices.

The place where there's the biggest impact is to have them in the universities. No university student should graduate without having used the PC and the Internet as a major tool for the work that they do. Ideally, you can get it down to very young kids, because it's amazing how good they are at learning these things and working with them, and coming up with ideas that no one else would about how to make it a better tool.

The costs of the PC are coming down. There are government programs now that are allocating resources to this. I think we have to think of it as starting a 10-year process to really get very, very widespread usage. Microsoft's role is to provide low-cost software for these educational projects, and to be a consultant to talk about what things are going on in the rest of the world, how those have gone well and, therefore, to provide any of our expertise to make sure that the money that is made available is very well spent.

MODERATOR: : Thank you, Bill.

Well, it's time for a break. I would like to say thanks everybody for your presence. And, Bill, thank you very much for your time.

(Applause)