Remarks by Bill Gates, Chairman and Chief Software Architect, Microsoft Corp.
Windows Hardware Engineering Conference (WinHEC) 2003
New Orleans, Louisiana
May 6, 2003

BILL GATES: Well, good morning. It's very exciting to be at WinHEC this year and talk about some of the advances taking place in our industry. This is an industry whose product improves faster than that in any other industry. And the number of different ways that people are using the PC also expand today that often surprises us and pushes us to move in new directions.

We are a very competitive industry, an industry where prices have come down as the power has increased at very amazing levels.

The penetration of the PC is continuing to rise around the world both in businesses and at home, particularly as broadband penetration skyrockets in a number of countries. In places like Korea, and now even in Japan, where broadband is expanding at the fastest rate, we're seeing the applications that will be pervasive around the world in the years to come.

I thought I'd checkpoint on exactly where the strengths are today and where some of the weaknesses are. Clearly, the economy has affected IT spending as a whole. The refresh cycles for lots of different systems are longer today than they've ever been. We can bring that back in somewhat by having enough improvements that we justify the expense, but there's work to be done there. Cycles in education are up at four to five years, in business three to four years and that's higher than it's been in the past.

Another thing I think is that we in some cases don't have applications taking full advantage of the new capabilities, and we'll be talking about how we get our industry more synchronized around these advances so that we deliver the full benefit of the improvements.

Perhaps the best example of that is the kind of graphics capabilities that are built into these systems and the fact that the standard interface of classic Windows applications right now is not taking advantage of those graphic capabilities.

With the next round of Windows, which we call Longhorn, this idea of the completely new presentation system and the standard user interface guideline requiring that rich graphics capability will take a leap forward and fully exploit the extra graphics capabilities. In fact, the way the experience will work is that people with even today's hardware will want to move up to the advanced graphics capabilities.

I've got three areas that I've marked as neutral where the industry is applying a lot of energy and I can see good progress being made: Trustworthy Computing, a product we recently released with Windows Server 2003 was a big step forward in that. That's our first major release since we went in and made the major focus on how we do our development methodology: code reviews and new testing procedures to focus on trustworthy as a top priority.

Interoperability: Here the future is based on the Web service protocols that take advantage of XML; great work going on but it won't be for the next year or two until we see that being used in a widespread way.

We see these Web services paradigms coming down to many levels of the system - not just being used across corporate boundaries for e-commerce, but between applications in the enterprise, even between the applications on a single system - and some of these paradigms will be even for system components. And so, advanced device driver models in the future will use those protocols to get more separation - more provability - of how that interface works together.

Robustness: A big advance here is having the feedback data on what the user experience is like, understanding from the Internet feedback information exactly where the pieces are not coming together. And particularly when the pieces don't fit together at all, when you have at some point a system crash now we have a complete picture of that, so what we're doing is expanding that diagnosis software to take in a variety of other situations, including the cases where the system thinks things are going well, but if the user is frustrated about the speed or the complexity of how they're doing things, we see that as feedback and are able to act on that to improve the hardware/software combination.

So, robustness. We'll talk about some examples where many of you have gotten engaged now in working with us to go through that feedback database and see for your piece, whether it's a driver or an application or a hardware design, what the problems are and how to make that better.

On the very positive side we have the growing popularity of Windows XP. We've made a faster transition to that new version of Windows than any previous version of Windows that we ever shipped. That's very important because of the richness, the robustness and getting the industry behind a single driver model, a single set of tests to take place. The installed base of so-called Windows 9x has dropped dramatically, and so investments are now focused on building and testing on Windows XP.

We've got hardware and software breakthroughs. I'll be talking about those. And we've got unbelievable price performance. The PC platform: now, there's no way anybody can say it's not the highest performance platform for any different type of application that they might be interested in.

Let's talk about the things that are changing the framework, the breakthroughs, that I think in the last year really proved to have more staying power, more energy than what's predicted.

Wi-Fi: Here we see in Intel a very major campaign promoting their Centrino brand, but really promoting their scenario that the PC can be anywhere, anywhere in the home, anywhere in the office, take it with you. The advances in 802.11, some of the things being worked out on hotspots now, Wi-Fi is going to be something we can assume in all the different computing environments, and certainly when we move up from B to G in terms of a compatible speed-up and then also increasingly to A where we get the even higher speeds.

So that drives a lot of new thinking that we have Wi-Fi. We need to do low power Wi-Fi. We need to get quality of service across Wi-Fi for things like voice connections so even in a portable phone, when it goes into your home, will roam to the Wi-Fi instead of using the wide area network, be connected in and get the advantage of the integration, and be able to send those voice bits across the broadband connection at lower cost.

The cameras and printers: This is the first year that even very demanding photographers are looking at the high-end digital cameras and saying, okay, we really are at the crossover point. People like Kodak and Canon and many others are building devices that are finally causing the change amongst those demanding users.

And in the printer area, people like HP and Epson, with new inks, new capabilities, are providing not only better colors, colors that work in all environments, but also printouts that last for many, many decades.

And so the scenario of photography alone will be driving home PC use, the number of hours there, the value of that PC.

For Microsoft's part, we think of things like photo story: in a sense, making it very easy to take a set of photos, to mail them, to pan and zoom them and be able to send them off to friends in a very rich presentation format that can be done even across a dial-up type connection.

Windows Media 9 Series is an add-on, an addition to Windows XP. We've had over ten million downloads of MovieMaker that let you build movies in a very rich way. We've had over 50 million downloads of Windows Media 9 Series itself. Now, we've got over 200 different devices, non-PC devices, that accept the Windows Media format and can play that back. And the quality, the compression there really enables a lot of things that weren't possible before.

We'll be talking a lot about screens, particularly TV screens. The size, the typical size of those screens are going up. The prices are going to come down pretty rapidly. With the ClearType capability and some of the things we're doing in our software, particularly Office 2003 with letting you read documents and page through them in a very natural way, that screen makes a big difference.

As we're getting these high DPI screens, managing the icon size and making sure that the window management works well even across that large display is requiring us to make improvements in Windows itself.

I mentioned Web services as the foundation for interoperability. The tools that make it easy to build those applications and use that architecture for systems management, for componentization as well as for high-level tasks like major application integration in e-commerce: the progress there is so rapid I would call it a breakthrough, faster than people would have expected.

And finally 64-bit: This year for the first time you'll see multiple 64-bit processor types running Windows and getting pretty fantastic performance. And this transition to support 64-bit on the servers and the high-end workstation, this is going to be a far simpler and smoother transition than the transition from 16 to 32-bit was. The way the tools work, the way the emulation capabilities are being improved, including new work that Intel is doing, we can make this something where you can make 64-bit and 32-bit systems, using the one that's appropriate for the things that you want to do.

This last year, we had a number of new ideas in terms of PC form factors. This is the first year that the Tablet PC has been out. It's been shipping about six months now. Those machines are largely sold out. If you want to get an HP or a Toshiba or an Acer or an NEC, the Gateway Motion machine - many of these machines - the demand has been such that the manufacturers have had to really ramp up the capacity they have, and we're getting lots of great feedback on those systems. Applications are doing a lot to take advantage of the pen and so we see that becoming mainstream in the years ahead.

The Media Center PC: This has been also very popular, the idea of the PC in the living room recording shows, having the TV guide, integrating where you've got your music, photos and videos together, projecting that out so that it's available to other PCs, a very great start there with the product that only came out this last Christmas.

Embedded Windows and Windows CE have also been going extremely well, whether in the form factor in the auto, in embedded devices or Smart Displays. We've been driving those forward and getting more and more sharing of the tools across all these different versions of Windows. So a lot of variety, in fact, more thinking about new form factors than we've ever had before.

So we think about this as a computing environment where the devices have to work together. When you get a cell phone, it doesn't replace your full-screen PC. When you get a watch, a SPOT watch-type device that will have information delivered to it, when you customize that to say what sports or stocks you're interested in, that should also work and update the other devices as well, so thinking of the user and making the transition across different devices very, very simple.

We see that as we get the new hardware we need new software and we need to drive that into full-blown experiences, things like what can the TV experience be like or handling family schedules or work schedules in a very rich way.

One thing we'll be talking a fair bit about today is the relationship between the PC and the phone. That's something that will be changing. When you get value-added on your PC without having to switch the phone that you use simply by having the PC be aware of what's going on, that integration we think is a very critical one and one that's influencing the PC hardware.

Wireless Bluetooth microphones we think will be very important, because they'll drive a lot of these scenarios, whether it's voice annotations, phone calls, phone calls where the screen is connected up as you're doing that, really new opportunities that leverage off the PC.

Looking ahead, the rapid hardware improvements will continue, whether it's the disk capacity that's so phenomenal and is bringing into play things like video storage and very high resolution picture storage, the increase in the graphics processors where the number of transistors now is almost as high as the processor, and in a few cases even more. It's pretty phenomenal, and it's really explained by the fact that we understand parallelism better in the graphics realm, those problems have many different units executing at once, we understand that better than in the general purpose code execution realm.

This idea of parallelization is becoming increasing important. In fact, as we see these clock rates moving up into the 10 and 20 or even 30 gigahertz range there is a big challenge where the performance will not necessarily continue at the same because of the huge challenges you get as you run up at those incredible clock speeds. There are some clever ideas to reduce leakage, to have advanced cooling systems that Intel and many others are pursuing.

For Microsoft, part of our contribution to this is to make sure that whenever something can be parallelized, we allow the tools to make it easy to express that, because having multiple cores running at very high speed is cheaper than trying to get up to those big clock rates.

So the same thing we've seen in graphics with the parallel capabilities exposed through DirectX, we now need that for more mainstream general-purpose code, and we have some very interesting ideas of how to take our CLR execution environment and have declarations that allow you see the parallelization opportunities and on the systems that have that we run very fast in parallel. If you don't have that, fine, you still get the same results; you just don't get the parallel execution.

Network speed is going up at a great rate. The gigabit Ethernet, that we see on servers over time coming down onto the desktop and then, of course, these wireless things being built into the system, something that we can just take for granted.

So fundamentally, the trajectory for the next four or five years is pretty solid. Even if it's not exactly the doubling rate we've had before it's certainly enough to enable things like speech and ink, very deep searching, much better user interface than we've ever had before, new scenarios including just all your digital memory stored and shared in a very effective way.

Well, if we look back, of course, there's been a lot of twists and turns in this industry as we've tried to take advantage of those rapid improvements. And we made a video that kind of captures some of the drama about the new systems, the new software and how those have come together so let's go ahead and take a look at that.

(Video segment.)

(Applause.)

BILL GATES: Many twists and turns as we've gotten this technology pulled together into systems that hundreds of millions of people are using.

One interesting question now is how we drive usability not just into the individual devices, but across the devices: the software interface being similar, the data moving between them automatically. And as you get these devices, you're navigating rich information: schedules, electronic mail, rich documents and the ideal would be that you'd have a way of doing that that is common.

We've put forward a proposal we call XEEL, X-E-E-L, which can be used on many of the devices. It's basically a roller that can be pushed in or moved to the left or right so it gives you four different directions and an action push, and then in some cases some additional buttons that let you switch in or out - back buttons, switch Window buttons - and we're proposing these for things like the Tablet PC or the remote control for Media Center or phone, PDA type devices so that we can get the user thinking, okay, I know how to move through electronic mail or schedules on each of these different things.

This is something that we'd love to see people trying out. It simply comes along with Windows, the right to use it and do it this way, and so I think this is key frontier is a similar user interaction across the different devices.

I mentioned the relationship to the phone as another area we see big changes. When a phone call comes in if you're at your PC, whether it's coming into your portable phone or your PBX phone, however it's coming in, you ought to immediately know from the caller ID who's calling, see e-mails from them, see the background from them. As you're talking go them, the ability to initiate a screen sharing session using a placeware type approach, that should be very automatic and running different applications editing information.

The ability to conference someone else in or take notes that are stored on your schedule with the actual call being marked on the calendar, that should be very straightforward.

Even the notion of when that call comes in of what should be forwarded to you if you're not there and how important it is to alert you that that call has taken place or simply taking the voice mail and moving it into the e-mail, many ways that the phone and PC can work better together.

We're using the network connected to the PBX as part of this. We're using the Bluetooth connection as part of this. And so in Windows, you're going to see a lot of enhancements that relate to real time communication.

Now, we've taken all of our thoughts about this future PC for the knowledge worker and worked together with Hewlett-Packard to put together a prototype that we call the Athens Prototype PC. We think it's suggestive of some interesting things and it shows how we're trying to get early prototypes for us to do the software work so that if these things catch on, if you're building them into devices the software will be there and users will get the full benefit of them.

I think the best way to understand this prototype is to actually take a look at it and see it in action and so I'd like to ask Chad Magendanz, one of our program managers for hardware innovation, to come out and give us a little demonstration of the advanced PC prototype. Welcome, Chad.

CHAD MAGENDANZ: Thanks, Bill. (Applause.)

Well, it's my pleasure to introduce you today to our vision of the future of business desktops, a collaborative effort between Microsoft and Hewlett-Packard, codenamed Athens.

Now, the most striking feature about Athens when you first walk up is the display: wide, high density, flat. It really has obvious consumer appeal, but we're seeing dramatic productivity benefits of a display of this type as well, both in our internal and third party research. For example, we're seeing a 30 percent improvement on time in task with large displays. We're also seeing a 17.4 percent improvement in reading speeds with high-density displays.

So there is some really productivity value in here for enterprises, but at what cost? Obviously a display like this would cost over US$2,000 right now. Fortunately, display research projections show that a 20-inch version of this same resolution display will have an average selling cost of less than $400 in the year 2004.

One of the things that's less obvious about the display is the work that's gone into consolidating a lot of the components and cables that normally clutter your desktop within the display itself. This one cable from the CPU, which contains high-speed USB and video, connects a slim form factor drive bay, USB speakers, slim and array microphone, camera and a Bluetooth transceiver that drives this rechargeable wireless keyboard that recharges right here in the base, wireless mouse and a cordless telephone handset.

By pulling the elements of the PC that the user normally interacts with into the display console, we make it easier to more independently support and update the CPU and the peripherals. We're really preserving the value of the hardware investment in the peripherals themselves.

In the process we can substitute the CPU with a dock so that laptops and Tablet users can achieve the same no-compromise desktop productivity experience and sacrifice nothing for their mobility.

Athens also represents some long-term investments we've been making in the Windows system architecture to improve power management and security. For example, when I'm not using this machine it normally powers down to this state. This standby state alone cuts power usage in the enterprise desktop by about 78 percent, which represents about a $95 per unit per year savings.

Now, just because it's in the standby state doesn't mean that it's not totally manageable. In fact, I can still use the phone and the notification lights I have here, so when I walk across the room returning from my meeting I see immediately whether I've got outstanding voice mails, e-mail or an upcoming appointment.

Now, just because I can walk up and have the machine ready to go in less than two seconds doesn't mean we're done yet. We have to streamline the authentication process. I have here a flash drive that actually has an integrated smart card chip and a biometric sensor so not only do I carry my documents around with me, but I carry them securely, and I insert it just like I would insert the key to my car. And the PC comes on automatically. I just press my thumb to the sensor and I've restored my working environment.

Now, a few notes: This is prototype software as well as hardware. Hewlett-Packard and Microsoft have been refining and designing Athens as a system, concurrently developing hardware and software and dramatically shortening the feedback loop to ensure that the end user is never the first system integration point.

These prototyping tools really help us exercise the hardware and software integration. So, for example, I know I could pick up a handset and make a phone call in the normal way, but with the PC added to the equation, I can use voice dialing or I can select a contact directly from the list.

But a broader goal is to work with a wide variety of telephony devices and telephone services, whether it's the analog phone that I've had in my basement for three years or an enterprise PBX system like this handset from ea.telecom or a Bluetooth solution like this off-the-shelf Sony Ericcson phone, this prototype Bluetooth smart phone or this prototype Pocket PC Phone Edition from Texas Instruments that not only integrates Bluetooth but Wi-Fi as well.

The goal here is to have a consistent user experience and a consistent set of application interfaces that really extract the complexity of the telephony integration solution from the user and from the application developer.

So I can't show you all of these integration scenarios in the keynote today but we have working demos of practically all of these up in the innovation room and I encourage you to stop by this afternoon to Room 265 and experience them for yourselves.

So let's take a look at this usage scenario. I'm going to go ahead and start up my music. Now, let's say I'm just working and I'm waiting for an incoming call. Well, an incoming call comes in and I can answer it with the speakerphone button right here on my keyboard. I could pick it up on the handset. "Hey, Mike, can I get back to you later? All right, thanks."

So let's kill my music and talk about what just happened there. The incoming call immediately muted my music so it wasn't a source of background noise for my call. It also set my presence information to indicate that I was on the phone. When I hit the speakerphone button it actually creates a virtual speakerphone from the speakers and high quality microphone that are already built into my system. When I picked up the handset, it automatically transferred the call to the handset and the same thing would have happened if I'd picked it up on a Bluetooth headset.

Now, the most important thing is this information page. Doing a reverse lookup on the caller ID I was able to get all my contact information from Outlook, plus I get my complete call history, including all my notes from previous meetings, mail and messages that I've exchanged, files that I've exchanged. This is the kind of information I need to be more effective in my communications as I'm on the phone.

Now, with all this information, we have a burden of responsibility to make sure the user stays in control of that flow. One way we do this is the same method you're probably already using at work; you shut your door or you put a yellow sticky on the outside of your cubicle saying, "Do not disturb" or "e-mail only". Well, we have the equivalent right here on the Athens PC in the "do not disturb" button right here on the keyboard. Not only does it light up here but it pops up over in the CPU to show that mode.

When that mode is in effect, incoming calls, as you can see right here, go immediately to voice mail so I don't have any interruptions in my workflow. I don't have the pop-ups, I don't have the audible alerts and my presence information is set to busy. I'm still able to screen the messages on my auxiliary display on my handset, but it really doesn't interrupt the flow.

Now that I have voice mail I can press the voice mail light on my keyboard and I can pop right into my unified inbox. Now, the unified inbox you're probably very familiar with. It has my e-mail messages, my faxes, my voice mail messages plus my log of all my real time communications. Here you can see the message I just received, and actually Bill just forwarded a message from Shane Robison, Executive VP and Chief Technology and Strategy Officer at HP. I know he has a message for the crowd today, so I'm going to go ahead and open that and play it for you now.

SHANE ROBISON: Hey Bill, heard you were unveiling the Athens PC at WinHEC. Thanks a lot. I think this is a great example of how together HP and Microsoft are creating whole new categories of technology and at the same time addressing real customer needs. What's really cool about the Athens PC is that it enables rich communications experiences that combines voice, video and text messaging in a simple, easy to use collaboration tool.

So for our business customers who need to communicate outside their immediate work environment, the Athens PC delivers a lot more value than just plain e-mail or the cell phone; you get an experience that's much closer than when you're working and communicating face-to-face. And just like the collaboration between HP and Microsoft with the Media Center PC, I think we have something really special here that will bring a lot of value to our customers and a whole new opportunity for our industry.

I look forward to hearing what the folks at WinHEC think about the product and we at HP are really excited about it. Talk to you soon.

CHAD MAGENDANZ: Now, because voice is a first-class citizen in the Athens environment I can actually reply to Bill's forwarded message with a voice mail just by clicking the record button on the keyboard.

"Bill, thanks for forwarding Shane's message." And I can play it back right here. "Bill, thanks for forwarding Shane's message" and send it on its way.

This is the kind of thing you could easily do on a Tablet PC between meetings and never have to touch the keyboard.

So what did we see here today? Athens is really delivering a simplified power management, cable reduction and intuitive system controls. We've integrated telephony so that we're adding value to your calls, surfacing information that's already available on your PC or on the Internet so that you can communicate more effectively.

Really Athens is a nexus for business communication, integrating all forms of messaging - voice, video, ink, text, recorded or real time - into one seamless communication and collaboration experience.

Thank you very much. (Applause.)

BILL GATES: One key thing to get these kind of improvements into the marketplace with the greatest impact is synchronizing the different activities, the work that goes on in the software, both Windows and the applications, the work that goes on in the hardware, and then the testing to make sure that it's got a simple way that people will be able to take advantage of it.

Historically, if you look at an initiative like USB, which over time has been a fantastic success, probably one of the best things that's ever come out of industry cooperation, even so it took longer to get the features coming together than it should have. USB 1.0 didn't have the system support. Then we had lots of problems with testing of pieces. And so it took quite a bit of time.

In this environment we don't want to see that type of time lag and so we're working hard to make sure that all these pieces come together, including the dialogue that takes place here at WinHEC.

We are building big advances into our tools for the types of software development that all of you do. .NET Compact Framework is now out in the market being used in lots of different types of devices - cars, set-top boxes, PDAs, phones - a lot of success there with the idea of the byte code that provides processor independence for those low-end devices, far more performance effective than the equivalent Java type capabilities.

With our C compiler, which is used to generate either native code or .NET, we've got a full commitment to the full ISO C++ in this new version.

We're also integrating in a lot more capabilities. For example, there's a thing called PREfast that's now available that actually goes through code and checks to find classes of errors that are typical, including things like buffer overrun type problems.

And so the same tools that we created in our security Trustworthy Computing push are now being made available for both applications and drivers.

We think that this tool level is an important area to invest. Being able to prove whether drivers can hang or not is something that we now have the capability to do. That was a very state of the art thing that even 20 years ago, people were talking about could we prove programs or not. Well, now in very special domains and one we focused on as the driver domain, we can prove lots of things and find defects, erase condition type defects that would be extremely difficult to find any other way.

We also are creating an environment where whenever you change a piece of code we know exactly which test cases might be impacted and so you can simply run those test cases immediately rather than having to go through and run all the different tests. And that for a large piece of code like Windows is a huge breakthrough. It means that we can take something like a QFE fix and test it for regression rapidly instead of saying that there's a two-week period, which is how long it would take us to run all those tests; so a lot happening at the tool level to improve quality and security.

I want to focus now on some changes we're making with Windows CE. Windows CE, as I said, is doing very well in all sorts of different device types, including some that always surprise us because it's out there for people to use as they see fit. We've increased the capabilities of Windows CE quite substantially. We have a new pricing capability where we've brought the standard unit, small number of unit price down for the core pieces to $3. We have volume discounts, of course, that go even beyond that. For systems that include more capabilities the price goes up but it's still quite a bit less than it was historically. And we've also made it so in lots of ways you can use it for free as you're doing development, as you're doing things in the educational environment or piloting things, there's no royalty required whatsoever.

So we see Windows CE as being a good example where there's lot of customer feedback. We want to drive it into new things and we've shown the flexibility to make that happen.

I mentioned earlier the importance of the feedback loop in driving quality for the entire PC ecosystem. This is something I can't emphasize enough. The benefit of actually having those report back system crashes, system hangs and over time even report back when things are slower, when things are frustrating for users, that really allows all the development resources in our industry to be prioritized in a much better way, knowing which drivers are having problems, knowing which applications have problems, really being able to take a numerical approach to improving things.

You know, historically if somebody said, yes, I'd like to rewrite parts of the driver, they might actually have compatibility problems if they did that, even if they felt they were improving the quality of that code.

Here you can drive your work based on the information, and so for what we're doing for driver creators is slicing the data so they see exactly the report back that relates to their work.

We've created a Web site so people can see that literally on a daily basis, see the impact of the work they're doing and if you're not participating in this analysis project we think that it's critical that you get involved into it.

Here's one example where somebody had a firewall driver that was in Windows and it was causing crashes and so at its peak 80,000 crashes. Well, as we did the analysis with them and helped them get a fix out, they were able to propagate that into virtually all the different systems that use their firewall driver and drive the level of crashes down almost now to zero, and they're committed to get that down to zero. And so, a huge impact by having the combination of the feedback loop and then the ability to do the update.

We saw this with a modem driver, a modem driver that had some problems that MSN and other products ran into. By seeing the analysis data, in this case hang data, we were able to get feedback and get that driver fixed and get the patches out through Windows Update to all of those systems.

And so it's a live ecosystem. For our corporate customers we'll be giving them a slice of this data and so they'll be able to see which of their corporate applications have problems or even to see which models of hardware typically are more reliable or less reliable on a daily statistical basis simply because of this mechanism that we've got there; and so a huge thing that's an industry-wide opportunity.

Now let me move to the data center and talk about how things will be changing there. Performance: We've got performance. So what is the cost that still remains? Well, the biggest costs are operational costs. And the whole complexity of taking things coming from the design to the development to its operational environment and being able to see where there's problems, being able to adjust things dynamically.

And so we talk about that as an IT lifecycle challenge where you really want to model the application and its performance requirements and have one model that tracks what you expected and what actually happened in a single place, so the designer, the developer, the operational people, the businesspeople see what's going on with that application in a common way.

Now, with Windows Server 2003, we shipped the built-in automatic deployment capabilities, so provisioning Windows Servers with different versions of software can be done across the network in a very automatic fashion. So getting that new server release out was a very key milestone in this initiative.

We call it the Dynamic Systems Initiative. I talked last year about our commitment to it. This year we're seeing it actually rolling out and something that there's special work at the hardware and software level to make sure you're taking advantage of this.

We've taken and schematized the system description. We've schematized the application description. The industry we hope will be distributing those schemas actually with their hardware and with their applications. So when it comes to the customer they simply can combine those things and describe how they want the application to execute, making it so they can do that statically or in a very dynamic way.

And so this innovation means that the quality of applications running in the data center will be much higher and yet the operational costs of managing that will be a lot less than it's been in the past.

And so this is a dream that's been talked about for a long time. It's now really happening: Lights out data center with different applications using resources as they need to. We're working with a lot of you on this to pull this together. The response from the customers has really been pretty incredible.

I'd like now to turn things over to show you some great examples of this and also we're going to show you some of the things going on with 64-bit Windows. So to start with the Dynamic Data Center let me ask Galen Hunt, who manages this work, to come out and show us how it really comes together and how it works. Welcome, Galen. (Applause.)

GALEN HUNT: Thanks, Bill.

Today I'd like to show a prototype of what we call the Dynamic Data Center. The Dynamic Data Center consists of Microsoft's automation and resource management software with a well-defined network topology of Windows certified hardware, including servers, storage and network devices.

The Dynamic Data Center software adds storage and network provisioning to the server provisioning capabilities that we're releasing in Automated Deployment Services for Windows Server 2003.

The Dynamic Data Center that I'm going to show you today was built for us by HP as part of Microsoft's Dynamic Systems Initiative. The hardware includes an integration of HP Storage Works arrays, HP's ProLiant Servers and HP's switches with firmware and resource providers developed for HP specifically for their hardware running within Microsoft's Dynamic Data Center.

Let's walk over and show you the console that I would use an IT administrator to manage my Dynamic Data Center.

On the left you'll see a list of the server and storage resources within my data center and with resource bars showing the percentage of total resources that have been allocated already within my data center. My IT staff has grouped my servers into classes based on their common usage within our data center. So, for instance, my BL20s have been classified as IIS Web servers and my BL40s have been classified as SQL Servers.

On the right we see a visualization of a deployment of the sales tracker application within my Dynamic Data Center. The sales tracker application is a typical example of a distributed line of business application.

I've just received a request from my HR unit asking for a new hardware deployment for a benefits application. Today without a Dynamic Data Center, at this point I would rack hardware, ask my networking team to pull cables and configure network switches, ask my storage team to configure SAN arrays and verify this entire topology was correct before handing it off to the HR unit. The elapsed time for this operation would be measured in days, and would involve hours and hours of operator activity, prone and subject to human error.

With the Dynamic Data Center, I'll do the same deployment in just minutes. The new deployment wizard walks me through the process of creating what we call a system definition model, a life XML blueprint of the new hardware deployment.

First I name the deployment. Then I select a topology from a predefined list of network topologies. I could create a custom topology specific for this application, but in this case the HR unit has requested a standard two-tier topology.

I add three servers, three IIS Web servers to the application, one SQL Server and 100 gigabytes of SAN storage.

I then select a V-LAN for this application. In this case I'm going to request a new V-LAN created specifically for this deployment. Alternatively I could have attached the servers to an existing V-LAN.

When I press finish, the system definition model for this new deployment is processed. The Dynamic Data Center software allocates servers from the pool of resources, initiates an OS deployment to those servers, configures the SAN array and can create the network topology by configuring the network switches. It's also updated the resource usages bars to show the change in allocation.

In a matter of minutes I've created a new network topology, including a new V-LAN. I've allocated and configured four servers. I've configured SAN storage. I've connected the four servers to the V-LAN and I've connected the storage to one of those servers.

The entire process was driven by the Dynamic Data Center software. No pulling cables, no racking hardware and no complex configuration processes.

Now, I'd like to show you how the Dynamic Data Center allows me to rapidly adjust to changes in business demand. It's the end of the quarter and our sales tracker application is experiencing its heaviest load. The Dynamic Data Center software has just alerted me that the sales tracker is overloaded. The system has already automatically extended sales tracker to the limits of its resource Service Level Agreement and is now asking me if I want to add additional resources to reduce the demand on the application.

I'm going to go ahead and add resources and I do so with an Add Resources deployment wizard, which is going to go in and modify, update the system definition model for this deployment.

I'm going to add three servers into the deployment. I'll add 100 gigabytes of storage and press finish. When I press finish, again the updated SDM, System Definition Model, is processed by the Dynamic Data Center software and additional servers are configured.

In a matter of minutes I've allocated three new servers, I've connected them to an existing V-LAN, I've created new storage and connected it to an existing server, all without physically touching any of the hardware.

As an IT administrator, the Dynamic Data Center allows me to easily manage a pool of servers, storage and network devices. The Dynamic Data Center drastically reduces the effort that I have to extend as an IT administrator to provide an agile IT resource to my business.

I've shown you how Microsoft and our partners are driving innovation in the agile data center. I'd like to now introduce Dave Ciuba, who will demonstrate how Microsoft and our partners are driving innovation in 64-bit computing. (Applause.)

DAVID CIUBA: Thanks, Galen.

Well, as you know Microsoft has been working on 64-bit technologies for quite a while and we've made some great progress recently with the availability of 64-bit versions of Windows Server 2003, SQL Server and an update to Windows XP 64-bit Edition, all for the Itanium II processor family.

We also recently announced the development of 64-bit versions of Windows for the AMD Opteron and Athlon 64 processors. Now, 64-bit computing has many benefits for our customers. On the server side databases and high-performance computing applications benefit from the large addressable memory space and on the desktop successful early adopters have been technical workstation customers in the areas of computer aided design and engineering, digital content creation and other scientific applications.

For the first demonstration I'm running a Hewlett-Packard ZX 6000 workstation using the Itanium II processor. And I'm running a 64-bit native satellite imaging application from Paragon Imaging. We're looking at a satellite image of the Los Angeles area that covers over 120 square miles at one-half meter resolution, meaning that each pixel of this image represents one square half meter.

Now, the image is over 67,000 pixels by 57,000 pixels. It's over 10.8 gigabytes in size. And I just want you to notice how effortlessly we can move around this 10 gigabyte image.

Well, I thought it would be fun to do a quick tour of the LA area, and I've set up a path here, which takes us on our tour and we're starting in the marina, head over to the beach. We can speed this up or slow it down, check out the boats here, head over to the airport. And we're going to end up at Hollywood Park, where I'm going to zoom in and just show you how easy it is to rotate this 10-gigabyte image.

64-bit Windows and the Itanium II workstation from Hewlett-Packard are ideally suited for this memory intensive application. For Paragon's customers the benefits are increased productivity and the ability to work with much larger, more detailed satellite images.

For the next demonstration, I'm going to change machines. I'm working on a pre-released version of 64-bit Windows using an AMD Opteron workstation, and I want to show how 64-bit computing is changing the face of digital content creation.

Do we have any "Star Wars" fans here in the audience? Raise your hands. Well, like many of you I'm a huge "Star Wars" fan, and I was very excited to find out how Jack Films is using 64-bit Windows. Jack Films is George Lucas' production company for the "Star Wars" prequels Episodes I, II and III. Using pre-visualization techniques, Jack Films' goal is to show how Episode 3 will look well prior to the filming beginning.

I'm running a 32-bit version of Maya from Alias Wave Front, and one of the key capabilities of 64-bit Windows is the ability to run 32-bit applications.

In the initial stages of the pre-visualization process artists create the 3D scenes using wire frame models like this one of R2D2. They then animate them, add textures, lighting and other effects to create the scenes that make up the movie.

Now, the next stage of the production process is to render or translate the 3D scene data into the realistic images that make up the final frames of the movie. I'm working with a 64-bit native version of Mental Rays from Mental Images. Now, for production cycles with short timelines there's nothing more important than saving time. Using 64-bit Windows, Jack Films is typically seeing their rendering times cut in half for memory-intensive processes. Now that's a real benefit that impacts everyone involved in the production.

Let's take a look at some of the quality of some "Star Wars" content using a 32-bit version of Windows Media Player 9 Series. (Video segment.) You know, George Lucas has certainly been a pioneer in digital cinematography and at Microsoft we're working hard to make Windows the premier platform to create and view digital media content.

I'd like to show a quote now that I think speaks for itself in terms of how excited Jack Films is about 64-bit Windows. And I know that I certainly am excited and now I can't wait to see "Star Wars Episode III."

In closing I tried to show a couple examples of how customers are being successful using 64-bit technologies today. In the future Microsoft, Intel, Hewlett-Packard and AMD are all very excited about extending the benefits of 64-bit computing to other parts of the enterprise, to information workers and to consumers.

Thanks and now I'd like to bring Bill back onstage. (Applause.)

BILL GATES: Well, now I get to talk about performance, and it's kind of fun to talk about this because over the last 10 years I've been sitting in meetings, looking at various benchmarks and seeing that other systems were faster. Price performance of the PC architectures had the lead in that for a long time, but in terms of pure absolute performance these systems have not been as good, whether it was the memory back plane, the processor speeds, software optimization, there was still work to be done.

Starting about three years ago though, we moved into the top 10 and particularly for clustered systems we moved in that case to the top.

We reached a huge milestone just this month with the release of some new benchmarks that we announced along with the introduction of Windows Server 2003.

So where does that put us? Where does it put PC hardware running the Windows Server operating system? Well, in fact, in all the different benchmarks for transaction speeds, the PC server is the fastest. There's price performance, and that one surprised people in terms of what we've done there. What would surprise people is the pure performance for every different level of CPU capability. It's the fastest overall.

And we're dying now to have people with UNIX-based systems or using Oracle or mainframes to have them come in and try and challenge these results, because the pace of improvements that we're on, the pace that drove us to be able to have the number one performance, that's not slowing down. It's a combination of hardware improvements, more parallel execution and very substantial software improvements.

If we look at that on an historical basis, we went from in 1996 with 4,000 transactions a minute to now the over 500,000 transactions a minute, so a very serious improvement of over a factor of 100 during that timeframe.

It's also interesting to think about this in terms of the absolute requirements. We are now with very few servers able to meet the most demanding transaction requirements and even adding in many levels of redundancy that the software has great support for that fail-over capability you're still talking about a very inexpensive system to run high-end transaction applications.

That's partly why you're seeing the focus now that the hardware and software costs are so low. We're attacking the operational costs and the development costs so that the overall cost of moving to new applications continues to go down at a very rapid rate.

And so a lot of this was the server release, a lot of it was what we've done in SQL, a lot of it's what's happened with our hardware partners, but moving ahead at a pretty phenomenal rate, partly driven by these 64-bit advances. Those were critical in getting to this level and so we expect that to become pretty pervasive at the server level and eventually move down beyond the desktop as well.

Let me now introduce a new capability that over a period of many years we think will be in all PCs. This capability we call Next Generation Secure Computing Base. We're talking about the details of this here at WinHEC for the first time. There's over 16 hours of breakouts to talk about what it means to have a system that, even as it runs arbitrary third-party code, you can make security guarantees about the cryptography and secrets that are kept inside the system.

This is a breakthrough. It's a breakthrough that will allow for privacy guarantees, will allow for document distribution control. It will allow PCs to be used for applications that they are not being used for today.

It will allow people who want to keep things secure to work across organizational boundaries and so collaborate together, whether it's corporations working together or security agencies working together. This is a very key technology.

Now, there are many pieces that have to come together here. We've got an industry group called the Trusted Computing Group, that Microsoft is a member of, that's driving the idea of the key information and how that gets into the system.

As you'll see in the breakouts there's work on the processor itself. There's work on the keyboard and video display. There's work that we need to do in Windows itself that will be part of the next major release of Windows, the Longhorn code base; so a lot of pieces that need to come together.

We do think that over time this will be a feature of all PCs, and so we encourage you to attend those breakouts and think about how you can benefit from the opportunity and help drive that initiative forward and it really does address some very key scenarios.

So we've covered a lot of ground today, talking about the breakthroughs, talking about the way we work together. I think there are some big takeaways I hope you get out of this: PC platforms becoming mainstream in the data center, taking the management complexity down in costs, really innovating, not just being there as a great performance block, but addressing some of the other challenges as well. We've got new scenarios, whether it's tablet or telephony, photography, and we need to keep driving those forward. The new form factors are going to be complementary. There will be a lot of those and the smaller devices with things like Windows CE will definitely proliferate.

And Microsoft's commitment is to keep working with partners, being smarter about that than we've ever been, and continuing to drive the PC out, drive the price so more and more people can have access and drive the capabilities up, so that literally billions of people have the PC as the most empowering tool they've ever had.

I think it's going to be exciting to tackle these new frontiers together. Thank you.

(Applause.)