Remarks by Bill Gates
Microsoft Corporation
CA WORLD '98
April 27, 1998
New Orleans

MR. GATES: Good evening. It's a great pleasure to be here. I wanted to talk tonight about the opportunities that companies have in the information age to create what I call a digital nervous system. What I mean by this is the idea of being more competitive by exchanging information using the tools of this age, the PC connected into your computer framework, and connected into the worldwide Internet.

When Microsoft was started well over 20 years ago -- and at that time there was a vision that computing could be very, very different -- there was the belief that the hardware power would grow so rapidly that the key limiting factor would be the availability of a software platform. And it's been amazing to see how this exponential improvement has proceeded year by year. Following a Moore's law rate of improvement, the computer today is over a million times as powerful at a constant price as it was 20 years ago.

And there's no doubt that in the next 20 years, that same rate of improvement will continue. Recently, we've seen things like 64-bit processors, Alpha and Merced, which will be a big advance in terms of performance. We've got huge storage capacity that's allowing us to not only have data warehouses, but to start to deal with video and audio. We've got better screen technologies that will eventually create a PC that's like a tablet we can carry around. We've got lots and lots of speed and great ways of clustering these systems together.

And so, there will be no limit in terms of what we can do with these machines. The key thing, though, is to step back and say, what is the business goal in using technology? Every company wants to manage its basic operations. They want to be able to immediately know what's going on with a customer and provide the best customer service. They want to be able to look at their planned events and see what they've done in the past, and share what they've learned with everybody inside the company, or even outside the company, who would be interested.

They want to look at their sales results and find patterns, dive in and see it by time, by product, which requires it being in an electronic form. And they want to get rid of the paperwork that slows things down, and get all the information easy to browse.

The economy is changing more rapidly than ever before, and so a system like this allows people to react to unplanned events, whether it's a currency change, change in the stock market, or a new competitive event.

Now, the basic building blocks for doing this are reasonably straightforward. In fact, companies are already making huge investments in the personal computer, the Internet connection, the management tools, electronic mail -- every company is moving in this direction. But I would say that there'll be a factor-of-10 difference in the benefit that the companies who do the best job receive versus the companies who just implement these things without really stepping back and thinking about how it changes the way they work.

In a company that's using these tools the right way, electronic mail will be common sense, not just for simple messages, but sending around presentations, sharing sales results or reactions to things that are going on. The plan for the management tools will span all of the systems and let them bring information together for simple, centralized control. Reaching out on the Internet, so that the kind of information sharing you've got inside the company is also available to partners and customers, is also a hallmark of a great strategy in using these increasingly powerful building blocks.

Microsoft's role is a very focused role. We are specialized in building platform software. We decided early in our history that the structure of the computer industry, when we were started, was not the right structure for rapid innovation. The idea of a single company doing everything simply wouldn't allow for the kind of R&D and specialization that we think characterizes a fast-moving business. And so, we've seen companies like Microsoft in platforms, Computer Associates in enterprise software, Intel in chips, Compaq in systems, these are the companies that have really driven forward over the last decade and defined the best-of-breed solutions.

For Microsoft there are really three things: Windows in all its flavors, Office, and BackOffice. Those products are over 80 percent of our revenues, and even looking out a decade into the future, I feel certain those will be our primary products. What's probably more interesting, though, is the initiative we're undertaking that span all of these products. These are the initiatives that, by listening to customers, we use to set priorities inside the company. The best example of this is our focus on the Internet. Soon after we launched Windows 95, I declared not only inside the company but for the world at large that Microsoft had some catching up to do, that we weren't supporting the Internet well enough, and that that would be a number one priority.

Over a two year period, we made a lot of progress building in the protocols, getting onto the standards committees, starting to think about the tough security issues, the issues that would allow video and audio to be used across these networks. And now, we're very much a full participant in building rich Internet software.

So that's an initiative that, although it hasn't come to a close, no longer needs to be the top priority that it has been. In fact, our key priority at this point is building manageability into our systems. This is because of the complexity that we need to hide, because of the cost of ownership that we need to reduce, because of the need to roll applications out on a broader scale than ever before.

Two other key initiatives are simplicity and scalability. That simplicity is a pretty obvious thing. We just shouldn't have as many concepts, we shouldn't have things like the registry, or all these file formats, or these utilities. I think everybody has got their favorite story of how it was hard to upgrade, or how it was hard to find their information.

We went through recently and looked at where we could simplify concepts, and we decided that we could do this in a very radical way. We could take all the search commands, the search command in help, the search command for files, the search command for messages, the search command for the Web, and pull those together. We decided the way you organize things we could pull into one metaphor. And so, we're going to be able to drastically reduce the number of commands.

Another step we took was to really review all of the error conditions that the system could get into. And I have to say, it was amazing to look at how cryptic some of these error messages are. Even I looked at them and said, what would I do if I got that message. My favorite one is this one here, it says, the DHCP client could not obtain an IP address. If you want to see DHCP messages in the future choose yes, otherwise choose no. Now, that last part is really helpful, where it's trying to remind me of the difference between yes and no. But, I still don't know what DHCP is and what happens if I say no. Is that a bad thing? Would I really love to see this message again some time? Would I be smarter the next time I see it to know what it means? It's just as cryptic as can be.

What it should say is that there's been a problem finding your network server, and ask if you expect to be able to find it, and engage you in troubleshooting exactly what's going on without exposing you to this terminology. So, literally, there are hundreds and hundreds of tasks we need to do to make this system simpler.

Now, the vision we've got, and I think the whole industry is pursuing, is to make these products simple enough that everyone in the world, whether it's at home or in business, will want to use the product. We've got to take the enterprise reliability that has characterized very large computer installations, and bring that to the servers we've got on the Internet, while taking simplicity at a whole new level and building that into the end devices.

We went out recently and took a little look at how tough it is when all these pieces come together, whether it's in the business environment or the home environment. And so, we've got a little video that shows how this works today. Let's go ahead and take a look at that.

(Video shown)

MR. GATES:Well, it looks like we've got a fair bit of work to do on the simplicity side. Let's move to scalability. What's interesting is that with the Web, the number of transactions will be far greater than ever before. And also the desire to find patterns in usage. To data mine all of those Web clicks, and to see which customers are responding in which way will be of increasing importance. And so we've got to take the most advanced techniques for scalability and build them into the lowest cost systems, so that it's simply a matter of adding new hardware to get the fault tolerance and linear scaling that systems will require.

Now, the PC hardware industry is taking on this challenge. A key element, of course, is bringing in N-way clustering. We have two-way clustering today, which gives us some fault tolerance, but we need to go all the way to the unlimited clustering that these systems demand. Already we're demonstrating how you can do a lot of transactions, a lot of mail, a lot of Web hits against the PC platform.

In fact, in all of these categories, except database performance, it's in the lead, and very soon it will take the lead in that way as well. The reason for this is that we get a number of factors coming together. We get the processors, speed improvement, which is doubling every two years. We get the use of more processors, going from a typical four-processor server now, to 8, 16, and even 32 processors. And then we get the improvement that comes in the software and the clustering. Here we're showing a typical database benchmark, and over the last two years, about a factor-of-seven improvement in the speed that we're able to provide.

Now, that kind of increase, a factor of seven over two years, we'll have no trouble achieving that in the next several years. That won't be the end of the road though, because there will be new applications coming along that push us even beyond that. But, a lot of progress. I thought one way to illustrate this would be to talk through, very quickly, a few customer examples. Saturn's a great example, where they're using messaging in a strategic way, not just electronic mail, but eliminating all their paper forms to manage quality, which is their number-one metric in what they do. So, a great example built on top of PC technology.

In terms of big Web hits, lots of popular Web servers now are using PC technology. NASDAQ is a great example, going from 8 million average hits a day, to 20 million on a big day, and doing that using off-the-shelf technology. Finally, in this area of database transactions, we have a lot of people now running enterprise applications on PC technology, whether it's SAP, the majority of whose sales are now on PC technology, or Baan, or PeopleSoft, or any of the others. Penzoil is a great example of somebody using SAP R-3, in the same way that Microsoft does, and running it purely in a PC environment. So lots of progress in scalability, but still quite a bit to do there.

Well, let's move on to the final initiative, which is manageability. This is one where our real wake-up call was a couple of years ago. When we sat down and looked at these cost-of-ownership numbers, and started to look at all the range of costs, application development, data center, communications cost, we realized that good software architecture could not only provide more business, but could reduce almost every one of those categories. One of the first things we did was sit down with our hardware partners, and talk about what could the PC hardware designs do. For example, how could we make it so that all the hardware was automatically detected, to make it so that you could remote boot, to make it so you could replace a machine without going to that machine at all, without having to recover any of the files that are on it, because we store those centrally?

We defined a more strict hardware definition, without the older busses, where you can't do this auto-detection and called that the NetPC. Those NetPC concepts you're starting to see in all the PCs that are out, even the low-cost PCs, the high-end PCs, and that has been a big success.

Another new flavor we defined is called the Windows-based terminal, where you have a device that's only about $500 and you actually run the applications on the server. It's the same applications, no rewriting, the same interface. In fact, you can change somebody from a terminal to a PC and all they'll notice is that because they have their own processor, things get a little bit faster. And so we gave people an easier way to get involved in the system.

The biggest initiatives, though, had to come on the software side. In the near term, we've put out our zero admin kit, but the fundamental advance, the big milestone for us, is the next major release of NT 5. There are a lot of pieces in here that I think are very important. The directory is key, because that lets you have user policies and manage things centrally. I think intelli-mirror is perhaps the most important thing. People were always saying to us that there was a dilemma between letting people have their files down on their desktop, where they were very efficient, where they had the control, where they could be portable, or having those files stored on the server.

The server had a drawback, because it would overload the network and slow things down. But the server had the advantage, because it was backed up every night. And users could replace machines, or go and use someone else's machine very, very easily. And so our challenge was to give people the best of both worlds. We've done this with intelli-mirror, by logically putting the information centrally, but making sure that you never store duplicate copies, and then caching most of the information down on the local disks, so 99 percent of the time you're fetching the information there.

We're also building in, in the management area, a standard console, built-in application deployment policies and instrumentation called WBEM, which allows the Real World Interface to do some exciting things. Let's take a look at the application deployment. I've asked Noury Bernard-Hasan, one of our product managers, to come and give us a look at how application deployment will look different in this new NT 5 world.

Good afternoon, Noury.

MR. HASAN:How are you. Nice to see you.

Clearly, application deployment and management is still one of the key tasks that administrators face. So what I thought I'd show you, fairly briefly this evening, are some of the new application services within Windows NT 5.0. If we start over here on the server, what we have is the Microsoft management console. And what I'm going to be doing is using this to manage my directory. So we'll see here that I have a domain called Zor , and if I expand this out, you'll see that I also have some other units, which are smaller administrative units, that we term organizational units. We click down onto the marketing unit, you will see there's also one called Windows Marketing, in which I am defined as a user.

Now, as the administrator, what I'd like to do is to deploy some applications, just to those users in the Windows marketing organizational unit. So the way I can do that is by using what we call a group policy. And a group policy is a new feature in Windows NT 5.0, which allows you to set system policies on large groups of users, rather than on a single-user basis.

So here what I will do is go ahead and edit the Windows policy on applications. And you can see here that we have a number of things we can do. We can set software policies, we can set scripts, power management, and so on.

However, what I'm interested in doing is application deployment. You can see here in the right-hand pane that we have no applications currently deployed.

MR. GATES:Okay.

MR. HASAN:Okay. So, let's go ahead and do some. So, if I now click new applications, what Windows does is brings up a list of all of the packages I have available to me. Now, a package in this context is really an object that contains the application plus everything Windows needs to know to be able to install it. And we provide all of the tools to create these packages within Windows.

I'll first select Cosmos Viewer, and you can see here it now gives me some more options. I can either issue this as an upgrade, I can publish it, I can assign it, or I can disable it. Now, with this technology, the notion of publishing an application essentially makes it available to users to install themselves, if they choose to do so. With an assigned application, this is now a mandatory application, which will appear on every user's desktop once it's been assigned.

What I'm going to do is, I will assign Cosmos Viewer, and you can see now what Windows is doing is building a distribution point on the server so that when users go and actually run this, it will install on their machines. And if you can see up here, you can see that the deployment status is that it's been assigned.

While I'm here, I'd also like to publish an application. So, what I will do for this instance, I'll pick Cosmo 1, and I'll say published. So, what this will do, this will make this available to users to optionally install, should they choose to. So, they have it.

What we've done is, we've seen very quickly how an administrator can go about making these applications available. So now, if we cut over to the client, let's see what happens over here. So, I have just to login. Okay, now bear in mind I've changed the profile on the policy for the Windows marketing users, of which I'm one. So, fingers crossed, Cosmo Viewer. There we are.

MR. GATES:Sometimes in demos, beta software doesn't work.

MR. HASAN:Surely, some mistake.

MR. GATES:There's Cosmo, that's great. So that was the one that was assigned to us.

MR. HASAN:That is correct.

MR. GATES:So it showed up in our programs menu.

MR. HASAN:That's right. Now, in this instance, although we haven't actually installed it, this is what was advertised to us, so the first time we run it, it actually goes out to the server and installs the software on my client for me. The next time I go and run Cosmo 1, it will just be there because it is installed. Okay, so there we have Cosmos Viewer. So, I'll close that down.

Now, if you remember, I also published an application. So, I want to step through the process for, if I'm a user, how I would go about doing that. For that, I go to my Control Panel, and if we look we see a new add/remove program Version 2. Now, this is an example of the way this is going to be implemented. This is not necessarily how it will be, but it helps to aid the demonstration.

You can see the options we have here, to add, repair, or remove applications. I'm going to click select add. And I can choose to do this from a compact disk, or a floppy, a corporate network, or indeed from the Internet. I'm going to select from the corporate network. And what Windows now does is, it goes out to enumerate my software distribution servers, and it comes up and says, yes, indeed, and as we saw before, Cosmo 1 is available for me to add. So what I will do is, I will click add on there, and then Windows installs this on my machine. And, again, the difference here is that this is an application that the administrator has made available to me. And this could be a subset or a short set. So, if it tells me that I've installed it, click okay. Next, finish. And I think we're done.

So, let me open up Cosmo 1 now, and there it is down at the bottom, Cosmo 1. So, what I'll do here is, I'll just quickly print a quick file, one of these useful graphics files.

MR. GATES:Wow.

MR. HASAN:Give it some color, say okay, and let's save this. Okay, now just to prove there's no sleight-of-hand here, I'm going to call -- It's about 18:41 p.m. I'm going to call this file 18:41 p.m., okay, and we're going to save that. Beautiful, done.

Now, the final scenario I wanted to show you this evening was, I think we've all be in the situation where we've run out of disk space, so we tend to just dive into the file system and delete big files. And I don't necessarily know what the outcome will be. I'm going to pretend I'm going to do that now. So, let's go into the Explorer, let's open up my C drive, and let's see what have we got here. We've got a number of things. I know, program files, that's always a good candidate. And what's this Cosmo 1 thing, I think I'll get rid of that. I don't think I need that. This disk space is more important to me.

So, I'll click delete. It's telling me to confirm that delete, and I'll say yes. Cosmo 1 is now gone. Well, what happens the next time I want to go and open that file that I just created. Well, if I double click on my documents, there it is 18:41 p.m., okay, now you know that today what would happen is, you'd get a dialog box that would say what do you want to open this with. Not very helpful.

If I double click on 18:41 p.m, this time Windows knows what the application was, and it goes back to the server and it repairs it because it knows there are pieces missing from it. And, again, this will apply really to all applications.

MR. GATES:So, no matter how messed up my local machine was, it used the server to automatically fix it up?

MR. HASAN:Exactly right, and that could apply to .dlls, to executables, to any piece of that application.

MR. GATES:Excellent. It's great to see that. Thanks.

(Applause)

MR. GATES:Well, Microsoft's strategy, because of our specialization, makes partnerships very, very critical. Some of our most famous partnerships include the work we do with Intel on new chip approaches, work we do with Compaq on both their desktop and server machines, and a number of other companies. Our relationship with other software companies really has been crucial to the popularity that Windows has enjoyed. And Computer Associates has been an excellent partner and a great example of our basic philosophy. Our themes are complementary, core competencies, innovation, and whenever we were looking for what makes sense to do, just go out and talk to our common customers and see what they're driving us towards.

There have been a lot of highlights over the years. And I'd say a big milestone for us was a couple of years ago, when Charles called me up and invited me to spend more time. I got a chance to come down here and speak just as NT was becoming important in the marketplace. And I heard a lot from customers about what we needed to do, and since that time we've come a long way.

Computer Associates has established a development lab up at Microsoft. We built in a framework, this WBEM, which means instead of them spending their time trying to gather the information, we offer the information, and they can focus on the heterogeneous application that presents that in rich ways.

Today's announcement is a very new thing, which has us shipping a major piece of Computer Associates code as an important piece of the NT 5 shipment. real world interface has some very neat things that show off the instrumentation that's build down into the system.

And so, I'd like to ask Tim Kraus, a marketing manager for Microsoft, to come up and give us a look, and let us understand exactly what we're getting here with the Real World Interface built into NT 5.

(Applause)

MR. GATES:Good afternoon, Tim.

MR. KRAUS:Hi, Bill. How are you? It's good to see you.

Bill, I'd like to introduce you to Anders Vinberg (sp) from CA, who is going to drive the spaceball for us today.

And, Anders, I promise the operating system won't fail, if you promise the application won't fail.

So, what we're looking at here is, we're taking a look at the trade show floor of the World Resource Center across the wall or in back of us, and you notice that there are lots of NT servers displayed here in this Real World Interface. You'll notice the two yellow balls indicate an area where there is a warning, or a potential failure at some point.

So, WBEM is aggregating the data in a repository and the Real World Interface is actually doing correlation against that data and identifying where there are potential areas for failure. So, we also have the opportunity to look at other views. As an example, we can take a look at specifically NT servers, or maybe exchange servers, potentially. Or better than that, getting away a little bit from the technology view, we can take a look at a view that's very, very specific to a particular person in your organization, maybe an analyst, or somebody who watches a business process.

So, we take a look at Frank's system as an example, and we see here in the Microsoft booth on the trade show floor, there is a warning area that we should be aware of, and we need to go take a look at that. We also have an opportunity to look at busy SQL servers on the floor. And if you take a look at that view, again, we go back to the technology, and it's specifically looking at Microsoft SQL servers. So, now what we'll do is, we'll go into the area of the technology center, and we want to take a look at why that warning light exists in that particular environment. So, we actually zoom down to the offending machine, and we see that it's a Windows NT server, and we go through the Windows NT server into an area that Computer Associates refers to as unispace. And the unispace is the general information about the machine, the configuration relative to software.

You can see on this particular machine, we've got NT and SQL server running along with SAP and WBEM data as well. So the machine has done some correlation and made a determination that there's a problem in the SQL server data. We want to be able to view that alert or those events using the WBEM technology and the Real World Interface. So, we can go on down into the WBEM data then and we can identify a lot of information that's been aggregated. And this gives you an idea about the two-dimensional view that's available for people. There's a lot of system data here, and sometimes it's helpful for you just to get a data dump of all of the alerts or events that have taken place, so that you can do work against that particular data.

But there's another way to view this that's a little bit more interesting. There's also a graphical way to take a look at this data, and we'll take a look at the performance metrics, and you see the two panes here. On the left-hand side you've got historical data and the graph indicates that the SQL server usage is way, way up. On the right-hand side, you've got real-time WBEM data that's being accumulated by the system and displayed, so that you can see what some of the potential problems are. And so, you can see that the synergy between the Real World Interface here and the Microsoft SQL server and Windows NT, for an administrator or manager to be able to display their data real-time and also to keep a historical view.

The important thing to note is that the same data that's available on this Windows NT system and this Real World Interface for Windows NT is also available for the enterprise manager in a heterogenous environment. So, this data is also rolling up to the Unicenter PMG real world console as well.

That's all we've got.

MR. GATES:That's fantastic. Thanks very much.

MR. KRAUS:Thank you, Bill.

(Applause)

MR. GATES: So where is this all headed? Well, certainly you can't pick up a newspaper or magazine today without seeing the latest milestone of somebody using the Internet to do something new and interesting, or the latest new company that's been founded around the Internet. It's really almost like a gold rush, people jumping in, people feeling like, boy, if they don't get out there, they might miss something. That's leading to a very rapid piece of innovation. At the heart of this, we're going to be able to connect all these machines together and I believe that with the hardware and software advances, we'll be able to make them far, far easier to use.

We, in the last eight years, have increased our pure research spending by a factor of 10. And today's it's a very substantial investment on our part. And that's aimed at building computers that see, listen, and learn. And this is not a far-fetched idea. The idea of computers talking to us or recognizing our speech has been around for over 20 years. Now, people at that time were over optimistic about how quickly it would take place, because they thought we could solve the problem simply by paying attention to the wave forms that were coming in. What they didn't know then, but we know now, is that if you just look at things at that level, the sound level, speech is very, very ambiguous. It's only because of common sense and context that people are able to figure out what's being said.

A good example is that our speech group calls itself the Wreck A Nice Beach Group, because if you say "wreck a nice beach" and "recognize speech," those sound exactly the same. And yet, if you say that to a computer and see that on the screen, see the wrong choice, you're going to think it's pretty dumb, because no one else would make that mistake. And so it's by having the learning, and seeing what a typical conversation is like, that the computer can start to move up to human capabilities in this area.

I'd be so bold as to say that 10 years from now every personal computer will have seeing, listening and learning. And we'll look back on the machines that we have today, with the keyboard as the only way of getting the data in, as very large and limited and wonder how people were able to work with those. The goal will be the same, to empower businesses, to have great digital nervous systems, for competitive advantage.

And at the individual level, to create a machine that people find makes their jobs more interesting, and makes them more effective. So there are some really exciting opportunities ahead: companies that are going to lead the way with digital nervous systems; a lot of progress that Microsoft will be making in its enterprise commitment, based on the guidance that you're providing to us; and a lot of excellent things that will be coming out of the work that Computer Associates and Microsoft are doing together.

Thanks very much for having me here today.

(Applause)

MR. WANG:Great job, thank you, thank you.

We're going to go and sit for a little while, and we're going to bring some chairs up here.

MR. GATES:Now, we get a little table.

MR. WANG:I thought that was the chair, it scared me.

MR. GATES:That's to keep us apart, I guess.

MR. WANG:Great. Okay. Thank you.

I guess the first question, and I know that it was asked at the senior management forum is, now that you are a big fan of big Bertha, how's your golf game?

MR. GATES:I haven't had much chance to improve it. In fact, the whole point of the big Bertha is to help players who aren't too good, I'm afraid. But, I look forward to maybe getting a few more outings this summer.

MR. WANG:Bill, for general home use, including entertainment, Internet and Microsoft Office apps, should I use NT 5.0, or Windows 98?

MR. GATES:I think the market for home users will stay overwhelmingly a market for Windows 98. A few years from now, we expect to have a replacement for Windows 98 that will be based on NT technology. And internally we sometimes talk about that as NT home, NT consumer. We haven't given it a name yet. But, there will come a point where we move that technology down, even into the home space. In the meantime, you'll find a minority of home users wanting to use NT there, but the rich features of NT really make it the best choice for the business marketplace. So our message is pretty clear now. For businesses, we're working with them to define a reasonable transition to get to NT, whereas for the home space, Windows 98 will probably be their best choice.

MR. WANG:Great. Now, this is a guilt one, and it's true. Why is this question being written on paper, and not being entered on a computer?

MR. GATES:I think that's a great question, because inside Microsoft I decided to ban all paper forms. And, you know, I found I had hundreds that had grown up over the years. We were a real bureaucracy. And, you know, we put all of our human resources things online, we put all of the different options that people have online, it was a real step forward. And something like this, I think in the future, we should make it all electronic mail based.

MR. WANG:Well, those who work for CA know that, because the question was asked from 12 to 2.

MR. GATES:Do you still have that policy?

MR. WANG:Yes, we do. But, we are using exchange.

MR. GATES:Thank you. Great.

MR. WANG:What is your vision for the enterprise environment 5 to 10 years from now?

MR. GATES:Well, there's no doubt that even taking a time frame like that, the enterprise environment will continue to be heterogenous. I think statistically, PC technology will make up an even higher percentage of it than it does today. But, inter-operating with all the other systems. A lot of people are starting to move application development, even though their data can stay through the gateway support that we and others have, stay on the existing system. Five to ten years from now, I'd hope that everyone can do lights-out management, that you can sit at any console, if you've got the right log-on privileges, see exactly what's going on, and be able to feel like for any system that needs the reliability, it's simply a matter of taking commodity hardware (Compaq may not call it that, but let's say standardized hardware) and adding the redundancy into it to give you the kind of fail-over that you'd like.

So that won't have to be super unusual, super expensive, or require you to go back and re-architect your application, the platform should provide that for you. And I honestly believe we can bring the cost of application development, the cost of supporting new systems way, way down. You know, we're already starting to see that, when people use the right Internet-based tools, but those aren't pervasive yet and there's a lot of fine tuning to be done. But 5 to 10 years from now, I think we'll be dramatically better off than we are right now.

MR. WANG:Okay. This looks very good. How do you view Unicenter TNG, in comparison to other enterprise management solutions, in respect to their ability to manage NT?

MR. GATES:I'm not sure I can remember the names of those other solutions right now.

(Applause)

MR. WANG:What are Microsoft's plans to bring mainframe reliability, availability and scalability to the Windows environment?

MR. GATES:Well, those are very, very key initiatives for us. One of the things we're doing is working with customers to really track very rigorously what's going on in their environment. You know, are there cases where they're having to reboot the system, where they shouldn't have to? Is that a design problem? What's going on there? Are they seeing the kind of reliability from these systems that they've gotten used to with mainframes and those things? So a lot of it's going to be really flogging it through, you know, getting the meantime between failures up by a factor of two, and then up by another factor of two.

We feel that the investment we're making in testing is a key part of this. We now we have more people who test Windows NT than do the development. And we have some new testing methodologies that are very helpful there. Also, by being a very high-volume product, all the users get the benefit of the incredible experience that's behind the product. But, there's still quite a bit to be done. And I think really, it's this reliability issue that's the one that's going to require the most cleverness.

Scalability, the plans are in place today. I don't see any problem there. But, reliability is one that is even more important to us, and one that's going to require all of our best work.

MR. WANG:How will SQL Server 7.0's own management integrate with Unicenter? We will have 2000 sites with NT Server, SQL Server and Unicenter.

MR. GATES:Well, this is a great example, where a customer really has a choice of how they want to do things. SQL server comes with some reasonably sophisticated administrative tools included with it. And so one choice would be to just use those. But, of course, that just let's you manage the SQL Server environment. Another choice is to take Unicenter and have that be your primary interface, but then look at the data that SQL Server is making available, through the WBEM interface. And if you see something that catches your attention, then you can use that inside Unicenter to trigger the SQL Server administrative tools to come up and let you see things in even more detail. So you can use either one in isolation, or probably most typically, certainly for that kind of a user, you would use them together.

MR. WANG:This is a cute one. Do you ever get to just go to the grocery store?

MR. GATES:Well, mostly late at night. I go in. I don't want to be like President Bush and not know what the latest scanning technology is. So I'm always going to come in just to check and make sure. I'm always stunned at, you know, how many products are available. And I know I'm not supposed to go and say I'm stunned by that, because you know, everybody else is going more often than I am. But, yes, in the last week I was in the grocery store.

MR. WANG:Okay. Here's the question then, what is the price for a half-gallon of milk?

(Laughter)

MR. GATES:Low-fat or skim? Yes, the price, that's a good question. I'll look next time I'm in.

MR. WANG:Last question, what do you believe is the value of the partnership that you have with CA?

MR. GATES:Well, there's a real distinct philosophy that we share with Computer Associates, which is the idea of really being great at some focused things, and we both compete with at least one company, and others, who think they can cover everything themselves. And I think, in the long run, with the approach we've taken we're going to benefit each other immensely. We're going to make sure there's more and more systems out there that people need to manage in incredible ways. And your engineers --