With Microsoft now pushing a raft of new technologies, the strategic goals of the Redmond, Washington-based giant are playing a larger role in the decisions of IS managers. Microsoft's Steve Ballmer talks to Michael Vizard of Infoworld.
Q: Microsoft as a company kind of personifies the fat client. So what is Microsoft going to do to reduce the cost of clients and reduce the size of those clients?
A: Well, we don't have a goal per se of reducing the size of the client. We don't have a goal to fight thin clients. I think the whole issue of the network client is really a frustration that people have with two things: People want PCs to be cheaper to manage. One magical-sounding approach to that is the thin client -- which might in fact do exactly what people want it to do. But the fundamental need that we think is being expressed is making these things much cheaper to own and manage. We're very focused on that and for the last six months or so that's become sort of priority two. If the Internet is priority one, this cost of ownership issue is really priority two.
Q: So are PCs too expensive?
A: Part of this issue is what is the cost of putting in a client. In a certain sense you can say that things get better, cheaper and faster every year, so how can that possibly be the issue? What would be the cheapest way to build a low-cost computer today? It would actually be to take a 486 [with] 4Mb of memory. Intel can build chips cheaper than anybody else can; memory costs the same no matter who it comes from; you either put a disk in or you don't put a disk in.
The real issue is people want the performance. If you look at what's happened just in the end-user community, the average price of a computer sold at retail was up Christmas '95 over Christmas '94 by about US$300. So if you just look at the average ticket at a place like CompUSA, their average machine price is up, not down. And why is it? People want to invest in something that's going to last for a longer period of time. Over time, I think what we have to do is to really tell a story that says "Look, there's going to be a great PC for US$800 to US$900."
I don't think it's going to be US$500 or US$600 in the foreseeable future, but I think it could be US$900 or US$800. And that PC is going to have software that will be exciting for the next three or four years. And that will mean an explicit set of development and design decisions from the hardware community.
Q: So why don't we already have that today?
A: You can't buy a small hard drive anymore. Try to go buy a hard drive today that only has 200Mb. Nobody will sell you one. Try to go out today and buy chips to build a 486 PC--nobody will sell you one. So the hardware community and the software community have got to pick a design point that continues to target that US$900 machine, that doesn't always target the latest and greatest.
You may have to have two development paths -- a high-end one and a low-end one. I just got a 20-page memo from Microsoft vice-president Nathan Myrhvold commenting about what such a device would mean, how would we manage Office so that we could continue to support these lower-end machines as well as higher-end machines. Lower-end machines are two years ago's high-end machines.
So the question is do you continue to enhance the old stuff or don't you. So when people talk about the Internet terminal, our No. 1 thing is to say "Look, there are two concrete frustrations that are being expressed here: manageability and initial price of the device. I think the core thing we've heard from our customers is focus in on those two issues, make sure that the PC does a good job.
There still may be a market for devices that are not PCs, that are thinner than PCs. And that's why we're doing so much work on wallet-sized devices and other thin devices. But I think the bulk of the market will remain PCs as it is today if we address the core frustrations our customers are expressing.
Q: In general, what kind of impact is the Internet having?
A: The number of companies that have really thought about how they want to re-engineer business processes on the Internet is surprisingly low. When I talk to VARs and other systems integrators, you'd be surprised -- probably 30% to 40% of them don't have Internet email addresses.
That's surprising, amazing, mind-blowing in a certain sense to me. But the level of total embrace of the Internet that we've seen has been somewhat lower than we might have guessed. In terms of customer scenarios of importance, the Internet comes up right away. With all the attention in the press we find that one of the top pressures today on IT people is business people who are saying "Oh my god! Are we going to be left behind? Are we going to get re-engineered out of business here?"
We did this announcement with WalMart with our Merchant Server effort, which they're partnering with us on. And it's partly out of the fear of being obsolete that companies even as big as WalMart say "Look, we've got to move now."
Q: How is Windows 95 doing in corporate sites?
A: The Windows 95 launch has certainly been an important thing for us. Win95 is now installed on between 70% and 75% of all new computers shipped around the world. And so it has gotten off to a very, very good start. People say, "How are the upgrades going?" We're above our plan for upgrade sales. I'd say probably below our plan in the largest businesses; not as much as sometimes people write, but probably a little bit below where we might have hoped to be in large businesses, but overall we're very pleased at the reception.
Q: How is NT faring?
A: The NT business and our server businesses have grown dramatically. It's up over 150% year-on-year. I don't know how people measure these things. We've got a server business that's almost a billion dollars a year. By comparison, Oracle's server business is about US$1.5 billion to US$1.7 billion; Novell's server business is about US$1.7 to US$1.8 billion. So we are now over half an Oracle or half a Novell in terms of server software sales.
The other big investment we've made as a company is in technical support. We're up now to over 2000 mission-critical, seven days a week, 24 hours a day support people around the world, which may not be what you can get from IBM, but it's actually more people than Oracle would have in that function. I think I'm prepared to say today that you can get a wider range of support offerings and better turn-around on mission-critical software issues from us, at least as we measure our performance, than you can from almost anybody in the industry with the possible exception of Hewlett-Packard, who we think still represents probably the state-of-the-art in terms of high quality customer support today.
Q: There's been a lot of issue about Windows 95 and NT. You seem to be pushing Windows 95 for the home and NT into big business, but people work in both environments. Is Microsoft ever going to synch up the code base for these two environments?
A: Let me give a few comments, because that is an issue that never got justice in any press. In a long-term sense, say eight years from now, we'll only have one code base. Everything will be on the NT code base. And you can take that as an absolute prediction for eight years from now.
In a long-term sense, I don't think our customers or us will we need to have a two-code basis. On the other hand, today we do. We need two-code basis for a variety of reasons. We have one code base that is smaller. It's more compatible, both from an application and a hardware perspective with what's come before it. And it happens to right now have a couple of features that the high-end doesn't, particularly in the area of plug and play, which is most germane for laptop users.
We have another system -- NT -- that is more scalable, has security built in, and is a little bit more demanding on the machine resources and less compatible, both from an applications and a hardware perspective. And you could say one is for "A" and one is for "B." Unfortunately, it tends to blend. We've found with our customers, people tend to go to NT Workstation when they're bringing in a new machine, which typically means a replacement machine.
What that can mean, of course, is in many worlds having three platforms: You've got 3.1 on whatever you haven't touched; you've got 95 on the systems that you have touched; and you've got NT on the new systems that you're bringing in.
Q: Doesn't that present a problem for IS?
A: It's less desirable than having two. I don't think you can ever have one because you never get rid of all the 3.1 on that day. For people that have longer replacement cycles, the discussion is almost always about 95. For people who are trying to do big new projects, where they know they're buying new systems, it's almost always an NT Workstation. Now the one other piece of data that's probably important for me to add is I do think that standard, a much higher percentage of all new machines in the next 12 months will come bundled with NT Workstation. That is, the hardware vendors will pre-install it as the default, particularly on Pentium Pro machines. We've had a lot of interest by the major hardware manufacturers in just making NT Workstation the default on their Pentium Pro machines.
Q: I guess you can say in one sense that DOS and Windows have kind of blended together now into Win95. When you say eight years for converging Win95 and NT, are you implying that will be the single platform? Or will they blend and there will be six other products and we will still have to deal with all the variations?
A: That question really is a very good one. Those two will blend. Will there be a need for a high-end compatible thing with NT? We don't anticipate that need. That is, we think we've got an architecture that works pretty well for that period of time, so there's not a need to have an "A" and an "A+." That doesn't mean there won't be other add-ons and utilities and things you could buy, but will we really have two different things that you can choose from. But I say that today, knowing full well that there still may be some issues that drive us to have such a product eight years from now. I just know we won't need both the 95 kernel and the NT kernel. We could need NT++, but I don't think so.
Q: What's the future of 16-bit applications?
A: Well, right now I would say we have modest plans for enhancing our 16-bit applications. We have modest plans for enhancing 16-bit Office. If you look at the work in Office that we'll bring out toward the end of this year, that work will go on strictly on the 32-bit platforms. A lot of people are doing some of their intranet and Internet work on the 16-bit platform, but not all of it. A lot of the innovation is happening only on the 32-bit platform. We don't have an aggressive enhancement planned today for 16-bit apps.
Q: Does that mean there will be a problem going forward where some users will be using some features in 32-bit Excel, and then try to share data with people on 16-bit platforms?
A: That's almost a separate issue because you're going to have that problem anyway. It's not an answer to say "Oh, we'll put all the new features back on the 16-bit and then things are fine." The No. 1 problem I think in most corporations is that even if you could upgrade every system, you can't do it instantaneously anyway. So the real question is is do we have a level of interoperability. And I think in the next release of Office we've probably done the best job we've ever done to try to address the issue of how do you save in the old format in the new release so that when you send mail or put something on a server you do have interoperability that people want. I don't think we've done a perfect job, but we know that's an area of great concern for the user base.
Q: All the conversations around the Internet seem to assume that bandwidth is going to be infinite and free at some point. So what has to happen in terms of networking to make all these visions happen?
A: Inside corporate accounts people today have fairly rich network infrastructures. But for the intranet case purely, I think things are pretty good and just getting better. For the remote site, I think PPTP is an important thing to look at and all that implies in terms of support from the router vendors, the hardware vendors, etc. For the Internet there's a lot of issues.
If you really want to have a great experience between your company and your users, you want to make sure that they have some kind of guarantee responsiveness out of the network. You want the consumer who gets involved with that to have a good experience and until something like RSVP or some other way of guaranteeing a certain quality of service comes about, I think there's a real issue for the Internet. Now we happen to be very involved with RSVP and working hard with the router vendors, Intel and Sun, to try to make some of those things happen.