Remarks by Bill Gates
Microsoft Corporation
Network Operations Executive Summit
November 12, 1997
Seattle, WA

MR. GATES: Good afternoon. I'd like to talk this afternoon about the opportunity we have together. Then I'll have quite a bit of time at the end for whatever kind of questions you have. Microsoft is talking a lot now about what is called the digital nervous system. And this is the idea that every enterprise, whether it's large or small will need to have its information in digital form, and be able to take advantage of that to streamline decision processes to draw more people to make decisions, whether it's people inside the company, or partners, or suppliers. That the digital approach is the best way to do that in the information age.

This is going to be what determines which companies do well, and which companies don't do well. And it applies across quite a range of information. The way you track customer events. Anyone in the company should be able to call up all the things that have happened with a particular customer and see that on their screen whenever it's appropriate. Things like sales planning or sales analysis shouldn't be just numbers on a piece of paper. They should be in a form where you can dive into detail. Where you can see anomalies, where you can mail around a view of the information to other people.

You shouldn't have a lot of paper forms in this vision. It's absolutely simple once you've got PCs on desktops connected together to get rid of paper forms, and to have the kind of accuracy and streamlining that the electronic system can provide. Perhaps most important is the way you use digital systems for unplanned events, surprises, where a competitor introduces a new product, or a project is delayed, you need to get everyone on your team collaborating to decide how to deal with that.

Now, a number of the elements that go into having a digital nervous system are already pretty commonplace. A PC on the desktop of the knowledge worker, and a reasonable internal LAN are big investments that companies are already making. But, in order to take advantage of those investments, companies need to step back and think about how they get their workers very comfortable with these systems using electronic mail every day as the primary means of staying in touch. And then, how do they get the bottom-up phenomena of people suggesting how this infrastructure can be used in a better way.

And what this leads to is incredible opportunity for people who run network infrastructures. The quality of network infrastructure is a big issue for Microsoft. When we look out into our crystal ball and consider the next decade, we know for sure that processor speed will not be a problem. Intel will be moving clock rates up to thousands of megahertz. We'll have Merced. Moore's law with exponential improvements guarantees we're going to have the kind of system we need to make computers that see, listen, speak, learn, giving them the power to allow them to be part of the natural environment.

If we look at all the other elements, storage size, screen quality, inexpensive, high-quality flat screens, we know those things will be in place. And so we see as a key limiting factor how much high speed network infrastructure is in place, because a lot of the scenarios I'm painting require a high-speed network in order for them to work. A lot of the collaboration with a product like NetMeeting requires quality of service for the video and audio. The ability for a small business to be hooked up to his accountant in talking through how are these transactions categorized, how are things going, without that account having to come visit that small business, it's all very dependent on the network.

So for network operators, the digital nervous system has two implications. One is in terms of running your own business and being efficient with operation support systems and business support systems. The second has to do with the huge market that this will provide for work at home, for consumer services, corporate services, and services out to the breadth of businesses that are fairly small and don't want to run their own servers, don't want to do their own backups, Web sites, mail servers, and all the things that the digital nervous system vision implies.

Microsoft fits into this by providing key software building blocks. We're going to continue to simply be a software company. That was the decision that was made with Microsoft got started, and at the time was kind of a wild decision because there was no market for independent software. Computer companies wrote their own operating systems. They wrote their own tools of all types. And it was only the revolution of the microprocessor that restructured the industry to allow for greatly increased specialization. I'm a big believer in this kind of specialization. Hiring great software people, managing them, managing the incredible testing methodologies that are required for these products. That is something that requires all of our attention. And so we don't choose to get into any businesses like chips or networks or systems, or those other things. Building Windows to be the operating system that has that natural interaction will be a monumental effort and require us to continue to increase R&D very dramatically in the years ahead.

In the world of software, here's a neat thing that comes with volume, and that is the ability to take software products that we put billions of dollars into developing and create versions of those that sell for under a hundred dollars. The magic of that volume is what's let us move forward aggressively, improving Windows more rapidly year after year. Ninety percent of Microsoft sales come from three products, Windows, Office and BackOffice. And looking out that 10-year period, I'd say 10 years from now, again, it will certainly be on the order of 90 percent of our sales will come from those three products.

Now, we have a number of initiatives that cut across these products. The initiative of the Internet support was one that we really declared very publicly in late 1995. We've come a long ways on that, getting the browser interface into the operating system, so the way that you look at local information and the way you look at remote information are now coming together. There's no reason the customer should think of those as being different.

On the Internet, we've built the highest speed Internet server into Windows-NT, and we're constantly extending that because the Internet itself is ever-changing. A few years ago, a great Web site was just a lot of static text. Well, today, a Web site is a lot of applications that run and present the opportunity to do transactions and present customized information to the user, and making it easy to build those sites, as every business on the planet will want to have a Web site, is very important.

There are specialized rules required here because most people won't want to host their Web site, and yet they have to be able to dial in and do the updates electronically without any manual effort there at all. And a lot of our tools are focused on that scenario. Where you, the network operator, will be providing the hosting.

Two other big initiatives for us are simplicity and scalability. Scalability speaks to the idea that the volume of transactions on the Internet will go far, far beyond what the world of computing has ever seen before. It's not just a question of matching what the mainframe has been able to do. But using new techniques involving component software and clustering to get several orders of magnitude above that.

We have just recently shipped our first version of clustering, and there's a lot of innovation taking place here that will let us meet the needs that the Internet will be bringing along. The performance improvements come both from the software advances as well as the chip advances, as well as the clustering and symmetric multiprocessing architectures. Those are key elements to make it all come together.

A final initiative is a focus on simplicity, and I'd say that is our top priority. We need to make it a lot easier to work with a personal computer. We've been looking at all the error messages in our products and saying, what would an intelligent user looking over the shoulder of that PC user, what would they say when that message comes up? What would they ask you to explore? And why can't that intelligence be built into the machine in such a way that in many cases that message shouldn't have to come up at all, and if it does have to come up, we should be able to guide the user along. There's a lot that we could do here to make the PC easier to use.

Part of our improvement here will come from taking advantage of the network. We want all the state of that PC to be easily visible from across the network. In fact, many people have raised the question of, should computing be re-centralized. Should all the processing and file storage that today is on the PC, should that be moved back to the center? Well, in fact, in a certain way, the answer is yes. We don't want to give up the responsiveness, the flexibility, the mobility of personal computing.

But with NT 5 we've come up with an approach called Intellimirror, which replicates all of the local state back to the server. And that gives you the best of both worlds. It means that for the files that have been copied down to your local disk, you get the speed, you can go offline. All the current applications work. But it also means that because the information is copied back up to the server, any time you're connected your files are backed up every night. If you go to someone else's machine, no matter where it is in the world, and get connected up to the Internet, your files are available to you all of the time. You can share that information very easily. And so it's the best of both worlds, the centralized computing world and the distributed computing world. We're doing that without asking the applications to be rewritten whatsoever. It is completely transparent to the applications that this is taking place.

When we talk about scalability, we're pushing the limit at both ends. At the high-end, it's the incredible emphasis we have on Windows NT. And it's not just the performance statistics that we measure ourselves by here. A key element is the robustness of the system. We need for Windows NT to be configurable so it's usable in carrier grade scenarios. And we have a lot of great partners who are helping us put those elements together. Down at the low-end, we've talked a lot at this event about Web TV, and how we see that bringing in new users, and starting to pull TV viewing in as an experience where you want to communicate while you're watching TV. You want to be able to interact with Web sites, you want to be able to do transactions. And so that simply adds to the number of devices that are connected to the Internet as a whole.

Another category that is very small today is the handheld, pocket-type machine. We're certainly optimistic that two-way digital wireless networks, although with somewhat lower bandwidth than the wired networks, will emerge at a low enough cost that it will be commonplace for all of us to have one of these pocket devices. And with the advances in handwriting recognition and speech recognition, being able to work with one of those devices will be far easier than it is today, at the same time as the price and battery life, screen quality, all the elements of the system will be improving dramatically.

And so Microsoft finds a lot of its R&D is on the very high end, to build the servers for the future of the Internet, and on the low end, to build the simple, pervasive devices that will help drive the Internet to its full potential. A lot of partnerships are involved in making this all come together. I'm sure this is a very incomplete list that we've got here.

In fact, I've just flown back from being at a Cisco event over the last couple of days, where John Chambers and I spoke to a group of their key enterprise accounts, talking about multimedia applications, the cooperation we have on the directory that you heard about earlier. And we certainly view that as a very key partnership. I've got a lot of others on here. And you can see, we have partners who are involved in DSL and in cable connectivity as well.

At Microsoft, we're very interested in working with anybody who can provide the high speed connectivity. And we hope there are all sorts of breakthroughs and good ideas about how to make that happen. Today, I hope you're getting a clear view that we are very optimistic about DSL. In fact, some of the new flavors of it, that are easy to deploy, we think, can probably move out and even get ahead of the penetration of other high speed connectivity.

We're also doing what we can to encourage the cable industry to move forward as well. Part of this idea is this concept of a next generation set top box, which can draw on a lot of the great work that was done with Web TV, the user interface, as well as the graphics capabilities that have been put together there.

Now, when people talk about simplicity, one of the key elements is making it easy to manage the device. This discussion about cost of ownership has become a very big thing, in the whole computer industry. When you really dig in to how people spend money on computing, there are many parts to it. There's the data center, application development, communications, and support and desktop hardware. And a challenge to our industry is to bring the cost of every one of those elements down, while at the same time demonstrating that the value of that system for digital nervous system type scenarios is greater and greater every year.

As we've really been focusing on lower cost of ownership, we've sat down with the key PC manufacturers and talked about what could we do to make this a lot simpler. And there are two big initiatives. One is called Net PC, and that's resulted in figuring out how to do remote booting across the network, how to make machines where everything in it can be automatically identified, and basically guarantee that no one needs to visit the PC in order to manage the software or hardware.

A lot of the Net PC advances are also applicable to the entire range of the PC family. So, for example, the remote booting will be a standard feature of all business PCs, starting next year, and the so-called PC 98 specification will be a requirement.

There's another approach that works well for light users, and that's to take the computation and move it to the center. For most users, this won't make sense, because the latency of the network, the desire to have offline capability means that they'll want their CPU locally, but we do offer the only thin client in the world, which is the Windows-based terminal. You can't be thin, as long as you're running the browser locally. The biggest application that runs on a personal computer, by far, is the browser. It requires the most memory, the most CPU, the most add-on. And so any device that can run a browser, the hardware will be as expensive and as complex as a modern PC.

And in the competition with the PC, anything that's incompatible with the PC has two big problems. One is that it doesn't benefit from the incredible volume economics that are behind the PC. The second problem is that by not running any of those applications, you're really taking away the key reason that people buy those machines. So in that class of device, the PC will continue to dominate.

People have been surprised to see over the last year that prices of PCs have come down. And that's partly, as we've gotten increased integration, as we've gotten the volumes up to about 100 million units a year, there is just a lot happening in hardware design that is bringing those costs down. Net PC type devices, simple ones, will be under $1,000. Consumer PCs have already moved under $800, and there will be machines that cost significantly less than that in the next couple of years. And so the thing to focus on now is making the management costs very, very low.

Another key element of that, of course, is the software. We've already brought out Systems Management Server and our Zero Admin Kit, to help with management. But, really tackling this in a breakthrough way has become the key theme of Windows NT 5. And so the IntelliMirror feature that I talked about, that replicates the server state to the client, and the client state to the server, is a breakthrough in terms of how people think about PC management.

Now, when you think about IntelliMirror, you usually get an image in your head of a local area network, in a corporation, setting up a server, and having their users connected up there. And that's probably the place that it will be used initially. But, there's an even more interesting image here, and that is where that server is actually not co-located with the machines, that the small business or the consumer has the machine at home, and that replication is taking place across your public network infrastructure.

The beauty of that is that with the right speeds there, you can guarantee that PC will always have the latest software, you never need to install anything locally, you can connect up on the network and ask for new applications to be brought down. If you have any questions, you've got enough bandwidth to have somebody come up on your screen, and through both with audio and video, talk with you. They can see the state of the machine, take charge, and be able to guarantee that satisfaction in usage is 100 times better than it is today, because you've got it connected up through the network, to a Windows NT 5 server.

We think this is a major milestone and it's part of some new services that we'll be rolling out as Windows NT 5 gets into the late beta tests, next year. The vision is simply managed Windows. Craig Mundie talked about how the Web TV is a managed device. Here we're talking about taking all PCs and being able to manage them. By managing them, your documents, your support, are all easily available through the network. And so the entire family of products, with Web TV, that uses Windows-CE, down at this end, all the way up to a server, say that is in a medium-sized business, that you remotely manage at a remote site, those things will come under this management approach.

So a key element of solving cost of ownership is to take advantage of the network, to turn it into the thing that really lets you always have the consistency and capabilities that you want there. And this requires collaboration between Microsoft and all of you, to make this available. We're already seeing some of this, for enterprise customers, but I think the biggest opportunity will come as it gets rolled out to a broader set of companies.

And so this is a time of great opportunity. We'd love to work with all of you on Windows NT as a platform for hosting a variety of services. We're getting a lot of third party support to build telecommunications capabilities on top of Windows NT. We see Windows NT momentum continuing to get stronger and stronger, partly because of our work, but also because of the things that our partners are doing. We're also very interested to hear from any of you how we can foster the investment in broadband networks. Are there things we need to do technically? Are there partnerships? Are there marketing activities? We view this as really the key thing that will determine how quickly we're able to deliver on the promise of the digital nervous system for businesses and consumers everywhere. And we're very excited about the chance to work with you.

Thank you.

MODERATOR: Now, we'll take some questions. I've spent a couple of days actually collecting questions that people wanted Bill to answer. So I thought we'd go ahead and begin with a few of those, and then open it up to some interactive Q&A.

MR. GATES:Great.

MODERATOR: So, the first question has to do with standards. The question is that Microsoft is often seen as building their own standards when cooperating with telcos and especially internationally. Will Microsoft commit to supporting standards, and what work are we doing there?

MR. GATES:Well, we're really putting a lot into standards because all these scenarios I'm talking about require an incredible number of standards. We do a lot in Windows to make sure it interoperates with existing systems, Unix systems, mainframe systems, and so whether the protocols to do that come from standards committees or whether they're other companies' protocols, we're very supportive of those, you know, SNA, the old NTP network protocols. We're very involved in the ITU, and that's been a new thing for us in the last couple of years.

There's a new organization called ETSI, which is a group that's looking at how the traditional phone network and the new IP networks should work together. And I think that's some very exciting work, and we're newly involved in ETSI. There's a video broadcast group called DVD that we're just joining that's going to do some great work. Two others that I'm very impressed with, W3C which is run out of MIT, they've controlled the evolution of HTML, and we've been a huge contributor to that. And now XML, the terminology is always very confusing, but XML is coming in to be very important. And then the IETF which plays such a central role on the security protocols, the directory protocols, the migration to ITB-6, standards that are going to let us get quality of service into these networks on a broad basis.

So there's a lot to be done in these groups. And I think if you look at it across the board, Microsoft is putting as much into these standards efforts as any company, and we'd be glad to know, you know, where are the new things. You know, there are a lot of standards committees in this world, and if we're missing one, let us know.

Standards are good for us. To the degree that there's more intelligent devices in this world connected to the network, it just grows our software opportunity. And, so we're -- you know, we should be the first company to jump onboard any initiative that's really going to improve system interoperability.

MODERATOR: Okay. Super. One of the things we've talked about a lot over the last couple of days is new network services, and how we are encouraging companies to produce those services to do it on top of our software. The question is, how we view these new services? Is it the view that we'll own the customer, or the network operator will own the customer, and how do we see that working?

MR. GATES:Okay. Well, Microsoft is a technology provider. We're putting together these building blocks. And when you use our building blocks to host systems, we don't know the customer. We don't have the customer's name. We're not talking directly to that customer. It's completely your relationship with the customer in terms of going out and provisioning those services. Now, that customer may be buying Microsoft Office from us. They may be purchasing our software. And so we need to make sure we're coordinated, that your services connect up to the client side software that we've got out there. But, the people on the communications infrastructure, they're the ones who have to guarantee the up time and the services that are at that end, and we will not be working with a customer. We'll simply be supplying you.

MODERATOR: Okay. Great. One other question that was gotten actually a few times is about your recent investments and rumored investments in the cable industry, and whether this means that we are somehow discounting the telephone companies, and technologies in particular like DSL?

MR. GATES:Well, we need to do a lot to make sure that people understand we are very neutral in terms of how this high-speed connectivity takes place. And we believe that there's a lot required to get DSL going. In the case of DSL, it's less of a financial thing about investing the capital, because the phone companies are structured a little bit differently, but I think you'll see us in a number of partnerships that relate to promoting DSL, and getting it to be concrete. The trial we did here with GTE that was talked about earlier was an incredible success. And now there are some new ideas about consumer splitter-less DSL that I think are fantastic.

So DSL is a bit of an untold story, and we need to get into some partnerships with some of the people in this room to get out there and tell that story. Now, we're not going to make our software only work with DSL. You know, we're going to make it work with the cable infrastructure and any of the approaches that come along, but people should understand, we are really anxious for creative ideas on how to drive this forward.

MODERATOR: Okay. Super. Well, let's go ahead and bring up the lights, and then if we can take some questions directly from the audience, that would be great. And I think we do have people available with mikes. We've got a question right down here.

QUESTION: Is Microsoft going to go to any special marketing effort to change the public's perception of security problems in regard to e-commerce?

MR. GATES:Well, we could have devoted this whole conference to security. It's a very important topic, and unfortunately it's gotten into the political realm as well, where the U.S. government has some desire not to let us export reasonable security software. And they may not even let us use reasonable security software domestically, the way things are going.

The biggest problem with security right now is how we authenticate individuals, the fact that we rely on passwords, passwords that people don't change very often. Passwords where people use the same password for very low security systems, and then that same password on a high security system. In the past, where vulnerability is quite dangerous because, unlike a credit card transaction where there is some ability to go back and protect the consumer by undoing that transaction, if your password is compromised, then your mail privacy is compromised. And there's no way to abort the transaction and put things back the way they were.

And so, we see the world moving to smart card slots built into all these devices, including the PC keyboard. We see a real need to get through these political barriers so that when you have a properly authenticated user, then you can make sure that the sessions are very, very secure that come from that user, and you can make sure all the operations up on the server are tagged properly so that only the people who are authorized are able to do the right thing.

We have made a bit of a breakthrough in terms of applying a single directory to all the different operations. We don't have one security system for files, one for Web pages, one for mail, one for database. The Windows NT directory spans all those things. So the likelihood that people actually keep the user group names, and the security privileges up to date is much higher when you get that kind of unification. But the weak link has to do with individual identification. There are some things that could be done right now to tell people that using a credit card across the Internet is no more dangerous than handing it to a waiter in a restaurant. It's kind of common sense in a way. And I think the word is getting out about that.

We were on a big site called Expedia, which is a travel reservation site, and we get so many users who come in and find out about flights but don't book them. And so we're going -- always going and asking them, well, what is the problem. And there is a portion of those where the horror stories about credit cards and the Internet is the thing that's holding them back. I think there need to be some industry-wide activities here, but the term "security" is broad enough that until we get people to use smart cards, until we get these political things solved, it would be inappropriate to go to the public and say that security broadly has been fixed and nobody should worry about it. Unfortunately, they should worry about certain aspects of security.

QUESTION: Do you envision the Web TV boxes or boxes based on the same technology becoming the dominant set-top box going forward? Are you positioning it like to be the standard set-top box?

MR. GATES:Yes.

(Laughter.)

MR. GATES:You know, now, I'm sure if you talk to someone else, they might say they have a set-top box they like. But we are talking the Web TV technology, the software and hardware technology and adapting that so it can be used as a set-top box in a cable TV network. And the actual price of that box is not substantially higher than a classic -- well, than just an MPEG II type set-top box, and yet it enables a number of additional services. And so we have just kicked off in the last few months discussions with people who connect to TV sets, including the cable industry, about, would they like to license that technology, and how might that work. And so we definitely see an opportunity there, and that is our only -- the only solution that we're proposing is based on Web TV and Windows-CE is the operating system.

QUESTION: You mentioned DSL and cable modem, but what's happened with satellite, I understand that you are in a very important project?

MR. GATES:Yes. There are other approaches. You might call them the dark horse approaches, power-line modulation, LMDS. There's a very special approach relating to satellites in low earth orbit, so-call LEO type satellites, and that's a company called Teledesic. But I'm personally an investor, and Microsoft is not directly involved in that project. But the idea is to launch about 288 satellites at about 900 miles high. And the idea there is that you could have high-speed digital connections.

Now, in no way does that compete with the terrestrial infrastructure because when it comes to serving places where you have enough population density that you can run the cable, or optic fiber, the satellite doesn't have enough capacity, because it can only serve about 15 people in a square mile. But when you get out to the rural areas where it's unlikely to be economic to run the fiber, even in the rural United States, and particularly in Africa, India, large parts of the world, you know, how are you going to get 3 megabit, 2A connectivity to a hospital in the middle of nowhere? This is where the satellite systems could play a very important complementary role.

Now, the problem with these systems is the fixed cost to get one launched is very high. If you just put up a few satellites and you say, yes, you know, 30 minutes a day you have connectivity, it's not very interesting. So, you have to get enough of them up there that basically you have 24-hour coverage. Teledesic is involved in building partnerships, raising money, proving out the technology, and that system should be launched in about the 2001-2002 time frame.

So, it does play a role, particularly for people who get out into the rural areas. I think that could be very interesting. But it's five years away. And it's not a direct competitor to cable or DSL, because it's where you can't do the wiring.

MODERATOR: The question, I'll go ahead and repeat it since there wasn't a microphone. The question has to do with outsourced services in the future, and who do we see being the dominant players in providing those services, whether it's companies like Digex and other smaller companies, independents that do that today, is it going to be the large telcos, or is it going to be entities like EDS or Anderson Consulting, and some of those?

MR. GATES:I think that's an excellent question. If you look at the marketplace today, the people who are doing the best job are the entrepreneurs like Digex who come in and base a company on the idea of hosting, and they figured out what it takes, how do you go about it? And, you know, this is complicated stuff. Companies like that have two choices, as these consulting organizations and telecommunications organizations start to get their act together and figure out that this is such a growth business that they're going to want to be in it, you'll have plenty of people who want to come along and buy your company and put that into their offering. Or you can continue to be independent, and just have the expertise to stay ahead.

That partly comes down to the question of, are there bundled offerings? Like when the phone company goes to the small business, they're not only saying we'll connect you up with DSL, but what they'd prefer to sell is a whole set of services where they manage the servers, they keep them backed up, so that they're a revenue opportunity for working with that small business which is dramatically higher. And there's no doubt they're going to use that customer relationship to get into those value-added services -- it's just too attractive for them not to do it. But certainly we'll be, in a sense, arming them, we'll be providing them with technology building blocks to help them do that. And, so there's going to be quite a bit of competition there.

Right now, I'd say that demand for doing that, for helping people set up Web sites, do remote hosting, demand exceeds supply. That the expertise is very limited. And so the employees that you have that understand this stuff, they're going to be highly sought after, unless they have unvested stock options or something, you had better watch over them very carefully.

But, in the long-run, 10 years from now, I think that's a very interesting question. I do think the phone companies are going to get a very big chunk of it, and the classy consulting companies, some of them will find it attractive and go after that. I can't cite any of them, though, where I'd say, wow, they've done a great job of this so far. People like BT may be the most aggressive, but they're very early into this thing.

MODERATOR: Bill, over here, how about right here?

QUESTION: I'm glad I get to ask this question. I've been trying for a while. And I think you've mentioned how manageability is an important part of your focus. And we provide a high bandwidth access services and network services, and network centric applications to small and medium-sized businesses over the last year, and really we're on the verge of kind of the intelligence of the network becoming greater and a lot of the intelligence, whether it be the quality of service mechanisms, or the RSVP type scenarios, and even its application to areas like active directory, services and profiles, and policies for connection oriented services are really moving to the N stations in terms of the initiation of those services, and the initiation of those qualities of service.

What are you doing and what are your plans relative to managing that environment? Typically, this goes in the base, the network vendors have given us the applications and network element managers that would help manage the hub switch's infrastructure, if you will, but as we get into more intelligent network services, clearly, a lot of these things fall in the N stations themselves. What are the futures, and what can you do for us, in terms of, relative to this space?

MR. GATES:Okay. There is an initiative, a very important one that Cisco and Microsoft are involved in, called WEBM, W-E-B-M. And that's the idea of having standard object descriptions and protocols, so you can get graphical views of all the elements in the network. And for many of the management tools, whether it's HP Openview, Computer Associates Unicenter, the management tools that are built-in to Windows-NT or the add-ons that we sell as our systems management server, they will support these WEBM protocols.

And so WEBM objects can include very high level things like an office document, or they can include a link and the status of the link, is it up, is it delivering at high bandwidth. And so as you're browsing through, saying okay, this user is complaining, and you're clicking on a variety of objects to see what's going on, you'll be mixing network type elements and high end software service type elements, in order to know what's going on there. And so, you know, we need some standards.

In terms of managing the Windows desktop, making sure the applications get installed, making sure you can remotely manage that, setting up the directory files for all that, we're going to build that in as a standard feature of Windows. You know, we recognized a year-and-a-half ago that we were woefully lacking in terms of the focus on that. And we've made immense progress. And that also gets in as part of what we'll ship with Windows NT 5.

So in your position, you'll continue to work with tools from us, as well as tools from classic systems management vendors. And all of the devices in the network will be creating events and exposing their properties, so you can see that information. And it's a very fruitful collaboration that's taking place between a lot of companies right now.

MODERATOR:Okay. How about there, way in the back?

Q Hi, Bill. I work for an RBOC that's responsible for bringing out the advanced, new DSL services. And one of the biggest inhibitors to rolling those services out is in the area of regulation and having to roll those services out under archaic federal regulations. What's Microsoft's position in terms of supporting bifurcating plain old telephone service, from advanced data services, and creating a deregulated environment to introduce those services in?

MR. GATES:Well, right now, our top priority is trying to avoid becoming regulated ourselves. And I don't envy you, being in a regulated environment. I think actually, that is a key reason why this DSL stuff, you know, people are still not pushing really hard, because there is this question of, should you do it in your regulated operation, should you do it in an unregulated operation. If you do it in the unregulated, how do you have to structure that, so that it meets some tests that are still pretty vague today? And if you do it a certain way, will it get tied up in the courts forever, you know, before you can really go out and get a critical mass of customers to make this successful.

And we certainly sympathize with that problem. And we'd be glad to be part of an industry-wide effort that would, you know, talk about cutting through the difficulties that exist there. It would be a crime if, you know, these DSL things don't get rolled out very aggressively in the next couple of years.

The interesting thing we might see, if the U.S. doesn't get the right frame work, is that DSL will be a huge hit outside the U.S. and held back here. In the computer industry, so many of the hot new things have happened first in the United States, that there aren’t even sort people paying attention to things outside the U.S. And, you know, we're seeing that in Singapore, Hong Kong, Stockholm, a number of places where the government created the right impetus that there are going to be high speed connections in a lot of these places outside the U.S., for millions of users, before we can say that's true, broadly, inside the U.S.

And so it's a challenge to us. It means we have to get our latest software out there, and out latest technicians out there, working in partnership with the people who have that infrastructure. It would be a lot easier for us if the U.S. was at least tied, if not in the lead, on getting the high speed connections available.

MODERATOR: Okay. Right here, please.

QHi, Bill. What is Microsoft's long-term strategy on Web TV, in terms of building global partnerships in the field of both sales and development?

MR. GATES:Okay. Our strategy for Web TV will be to pick partners in each country. We've made the decision not to just go on our own, in terms of doing Web TV outside the United States. In the case where the Web TV is hooking up across the phone lines, there are some fairly natural partners for that. To the degree it's hooking up across some other infrastructure, you know, there are people who it's fairly natural for us to work with on that.

We're just at the very beginning of starting those discussions. And so in the larger countries, just in the last month, we've put the teams together here that will be having those discussions with partners. Now, Web TV is something that, in order to make sense, you've got to get it to critical mass. You've got to get lots, and lots of systems out there, which means you've got to get retail distribution, you've got to get word of mouth, and it's a fairly tough bootstrap. So we are looking for partners who see the potential that we do and can help out with some of the infrastructure elements.

MODERATOR: Okay. How about right here.

QHi. How much do you think will it take until electronic commerce and electronic media get at least 50 percent of the total media and commerce markets, how many years?

MR. GATES:Well, in a sense there are two types of electronic commerce. There's commerce between a buyer and seller, that would have taken place anyway, and instead of using paperwork to confirm that transaction, they happen to use the Internet to do that. That's kind of trivial, and you can drive the numbers for that up, just very rapidly. All you have to have is, you know, Wal-Mart, using ADI more, people buying from Cisco more, you know, typical transactions, just moving into this form.

The thing that's much more interesting is when the Internet is allowing buyers and sellers who never would have found each other in the past, to be able to reach out and find each other, to find some service offering that's available in a different country, where you can collaborate across the Internet or find some obscure gift that you wouldn't have found in some other way. And that's a little bit harder to get at, numerically.

So what you're going to see in the next two years, with these big numbers, you know, e-commerce grows to N billion dollars, with X percent growth, but most of that will be the trivial switching, which I have nothing against. But, there is nothing profound about it. I mean, what if they switched to using a telegraph, would that be a big deal. It's the whole notion of the marketplace, the matching function that makes this thing a very big deal.

My biggest prediction on this is that in a decade's time, a significant majority of Americans will be living what I call a Web lifestyle. That is, they will turn to the Web four or five times a day, without thinking about it, without thinking that it's a big deal, in the same way that they live a phone lifestyle today. And they don't think, wow, I'm using the phone, that's cool. I'm really good at using this phone, my friends aren't quite as good.

And so you'll just say, I'm considering buying a new car. Well, certainly you may not do the transaction on the Internet, but you go out there to get the information. You want to book a trip, you have a hobby that you want to see what's going on. There was a sports score last night for some obscure team, you want to do something with that. And, you know, we're pioneering some of these types of Internet sites to learn more about that. So I think, you know, we're on our way, that people are used to electronic mail, people are willing to schedule their services across the Internet, that year by year, that's just going to get bigger and bigger.

Now, this -- in this metric, the U.S. is a little bit ahead of other countries, because everyone who comes out of college pretty much has been so exposed and is so used so used to it. And if you go to colleges outside the United States it's stunning to me that outside of the computer science department it's not nearly as commonplace. And so you don't have the same evangelization taking place by the young people who are moving out into the work force, and telling everybody that they've got to get connected up.

MODERATOR:Okay. How about there, way in the back.

QI represent a few Internet-Intranet start up companies, anywhere from 5 to 100 people in Minneapolis. I have this hypothetical question. Say, Bill, if you are starting your own business right now, you are a nobody, and you have several employees, and you might have $10,000 in the bank, and tons of credit cards. And there are 100 Microsoft solution providers down the street and you are one of them, how would you start up your business right now, in this highly competitive and highly exciting market?

MR. GATES:Well, the key thing is supplying something that you do better than the other guys. And you have to find the customer who is not being served in an appropriate way. And in some areas, like understanding Windows NT and really managing Windows NT networks, we are in a period, and will continue to be in a period for certainly three or four years, where demand exceeds supply. So if you're just a very high quality operation, you've got good, very well trained people, you're retaining those people, believe me, the Microsoft field rep in your area will be able to guide business to you.

Now, you want to have a strategy that lasts beyond the period where demand exceeds supply. During that period, you want to have built up an expertise and a set of customer relationships that will make you unique. And that's where you get into the question of specialization. Given that there are all these customers you can work for, why not pick customers of a certain type, so you can start to reuse more of your code, so you can start to be a lot better than anybody else is at dealing with that type of customer.

So finding your own path and something that you enjoy doing, I think there is a lot to be said for that, will mean that even when you get out of this period you'll have a business that will last.

MODERATOR:Okay. Here in the center.

QYes, Windows NT 5.0 and its associated technologies is a major undertaking, even by Microsoft standards. Do you realistically believe you can deliver this technology by mid-'98 to late-'98?

MR. GATES:Well, the answer is that we're choosing to prioritize the quality of the product. And given that that's the top priority, it does make the schedule uncertain. I'd say, I think it's very likely we will deliver the product during 1998. We feel pretty good about that. But if it comes to a choice where we're getting feedback during the beta tests, where people say, the scale -- make sure the directory scales better here, make sure this interoperability with UNIX works better there, make sure that the upgrade scenario is simpler in this way, then we -- you know, if we have to take extra time to get those things right, we will.

And the beta one product is out there. It's got a very high percentage of the functionality. The beta two product, that will complete all the functionality will be out, probably, I think, March-April time frame. And you'll just have to track along with us, you know, see what the feedback we're getting, see how people look at this product. I feel pretty good about this being a very, very important 1998 product.

But, you know, it is, as you say, the largest operating system undertaking that anyone has ever done. There are literally tens of millions of lines of code that are part of that system. And the number of hardware configurations, network configurations that this thing has got to work in, super, super reliably, is very, very large.

Microsoft actually spends more man-hours on testing than we do on development. That is the characteristic of large-scale, world-class engineering projects, is the testing is the hard part. And if anybody can start a company, and make a breakthrough in software testing then, you know, they'll make a fortune. There's a real need there.

MODERATOR:Okay. Any questions this side of the room? Yes, right here? No, go ahead?

MR. KLAUS: My name is Yahim Klaus from Deutsche Telecom. I would like to know whether you have set up some kind of a project to overcome the Year 2,000 problem. We have a lot of difficulties with legacy software, which is in place for many years, even at a time where we have, let's say, saved space in memory, in order to use just two-digit years. Now, we are running into some difficulties. Have you any programs, or can you assure us that your software is safe for the year 2,000.

MR. GATES:Yes, Microsoft has a fair bit of information up on its Web site, about Year 2,000 issues, where we've really dug into those. The simple summary is that, PC software, because it was written, you know, in the last 20 years, we're not using BCB representations for dates. And so, you know, Excel, Word, Windows, we don't have the Year 2,000 problem. And it's extremely rare for people who wrote applications on top of Windows to have reverted to sort of a punch-card storage format. It's possible for them to do so.

In terms of Microsoft itself, you know, we use SAP software running on Windows NT with SQL Server, and so that's already been set up for Year 2,000 and so there is no difficulty at all there.

MR. KLAUS: So just forgive the short comment, you are assuming that Windows 3.11 will never be somewhere in operation or 3.1, because I have -- (inaudible) -- that 3.1 you cannot go across the Year 2,000, it jumps back to somewhat in 1980.

MR. GATES:No, I mean, we should talk to you about that specific case. I mean, there is something very strange if you think there is a problem, because we don't store our dates that way at all. And we'd love to get to the bottom of what you're seeing there. So we should have somebody follow up.

MODERATOR:Well, let's take two more questions. How about over here.

QOver the last 18 months or so, we've seen increasing cooperation between Cisco and Microsoft, and it appears that you share several key strategic elements going forward, you mentioned WEBM, the various protocols, directory structures and that sort of thing. Based on your shared strategy and perhaps destiny, if you look out, say, two to three years, what kind of business relationship do you see between your two companies and what kinds of products and services do you think the two companies will be taking to market at that point?

MR. GATES:Well, I think there are three areas that we believe in. One is this whole directory area, because the directory is key. If you want to have rich security, if you want to qualify service levels by users, you've got to have the directory. So that's a big initiative that we're involved in. The directory also fits into the second area, which is cost of ownership. The WEBM management protocols are very key there, and so I'd see us as really telling a strong cost of management story.

And then, finally, multimedia applications. If somebody wants to do video-enabled NetMeeting, if they've fallen in love with that, what does it take for that to work? The way that networks are configured inside companies today, and certainly public networks, it's not always easy to do that. If you want to use natural multi-casting, it's not very easy to do that. And so, in order to make those kinds of technologies commonplace, we're both going to have to go to industry standards bodies and make sure that certain things get adopted on a widespread basis. And we're still going to be, you know, two different companies.

I think there is a good analogy between, you know, Intel and Microsoft. We operate as two independent companies but there are a lot of initiatives that we think are good for the industry as a whole. And as major players in the industry, we can drive those things. If somebody makes faster chips than Intel, I'm going to write my software for them. If somebody writes a more popular operating system than Windows, Intel is going to work with that guy. So we're not -- we're not coupled together, except to drive these initiatives that grow the industry.

MODERATOR:Okay. Last question, right here.

QWe run a Web forum, and a lot of times, with Internet NT we've gone to a new level, just as you have Enterprise NT, you have SMS, which is a reactive tool. We need a proactive tool, and I've heard Unicenter and NIQ and Tivoli mentioned here. But, we need something that monitors end to end. For example, if we restart Martha Stewart, I have to make sure her shopping cart is up and running, even though commerce server says that the service is running. So do you have any plans for a monitoring tool that will be proactive, that performs end to end testing?

MR. GATES:Well, we'll be doing some work ourselves in that area. This really opens up -- it's addressing an area where we haven't done enough, and we're really learning a lot about it. I called it data center tools but, you know, you're sort of a data center to the max. And you've got to know right away when something is not working, you can't wait until you go in and pull these things.

Good management software has always had the concept of automatic notification of things that aren't running. And so getting the third party vendors, the Tivolis, Unicenters, Openviews, to tie their things into RMS, and enhancing our own network software, we're going to have to pursue both of those paths in order to close the gap here.

You're a pioneer in terms of doing a lot of these things with Windows NT. And so, for better or for worse, we are learning a lot in terms of what do we need to do to our software to meet your requirements.

MODERATOR:Okay. Great. Thank you very much. I think we'll call it Q and A there, and we'll have Pete Higgins come up and talk about our content strategy.

MR. GATES:Thanks.