1988 Interview Between Scott McNealy of Sun Microsystems and MicroTimes
Scott McNealy talks about Sun Microsystems, RISC, UNIX, and NeXT
from the Mid-December 1988 issue of MicroTimes
The World According to Sun
President/CEO Scott McNealy Sets The Record Straight
By Mary Eisenhart and Steve Guttman
A typical problem that people have when they work with other people is that they’re afraid that people will steal part of their revenue. It’s sort of like being a little kid and wanting to eat everything and not sharing. But that’s how you make friends, and friends are useful, in that business is really done on the basis of personal and business relationships that work well. —Bill Joy
There has been a severe “disconnect” between what computer customers want and what most computer vendors are willing to sell, creating an extraordinary opportunity for Sun and others to respond.
With 99 percent of the general-purpose computer market still untouched by Sun, our employees recognize that far greater challenges lie ahead. —1988 Annual Report, Sun Microsystems
It would be easy to accuse Sun Microsystems of delusions of grandeur, if it weren’t for its impressive track record. It would be easy to accuse its leaders of arrogance, considering their penchant for taking potshots at the rest of the industry, if it weren’t for the fact that they’re often merely pointing out the emperor’s unclothed state. The sweeping statements to which Sun executives are so partial suggest less a company on a mission from God than a company which can’t quite figure out why the rest of the industry doesn’t do something which obviously works. Especially since Sun people seem to be having so much fun all the way to the bank...
In 1982, Sun Microsystems was founded with $4 million in venture capital by hardware innovator Andreas Bechtolsheim (whose work on the Stanford University Network gave the company its name), software wizard Bill Joy (one of the primary architects of Berkeley UNIX), Vinod Khosla and Scott McNealy. Their plan, as Joy told MicroTimes last year (Issue #34, August 1987) was “to build a full-fledged computer company based on microprocessor technology.”
This year Sun, which has roughly doubled its earnings every year of its existence, had its first billion-dollar year. Its phenomenal growth has been spurred by an obsession with technical excellence coupled with fundamental principles: adherence to industry standards, open architecture and cooperation with outside vendors. Part of Sun’s commitment to openness and strategic alliances undoubtedly arose from the ingrained “Live Free Or Die” philosophy of UNIX itself, which came into existence in response to monolithic computer companies who locked customers into using proprietary hardware and software from the same vendors; under UNIX, if you didn’t like the job Company A’s machine was doing for your business, it was an easy matter to buy hardware from Company B which ran the same software.
Another part arose from what Sun regularly insists is no more than common sense. Any Sun presentation is likely to culminate with a variant on standard Sun themes: companies dependent on proprietary technology are doomed; adherence to industry-wide standards and cooperation among vendors insures prosperity for all concerned.
Not only is Sun currently experiencing megagrowth, it’s spent the last year positioning itself for an even more spectacular future. In the connectivity area, it’s continued to develop and support TOPS networking technology, which it acquired last year and which allows Macs, PCs and UNIX machines to exchange files on a network. Following last year’s dramatic price reduction of its 3/50 workstation, which had the effect of giving a whole new range of options to the would-be Mac II or 386 buyer, it introduced the phenomenally successful 386i, which runs DOS applications in a window within a UNIX environment. The 386i was also a first for Sun in that it was based on Intel’s 80386 microprocessor, in contrast to its historically Motorolabased product line.
Sun also introduced a line of high-end workstations, the Sun-4 series, based on its own implementation of RISC (Reduced Instruction Set Computing) technology. Called SPARC (Scalable Process ARChitecture), the standard has been adopted by several leading chip manufacturers and stands a good chance of becoming predominant in the RISC arena.
In what may be the most significant move of all, Sun also announced a major strategic alliance with AT&T, under which the two dominant varieties of UNIX, Berkeley and AT&T’s System V, would be united in a single standard to be known as V.4. Further, in a move perceived as a direct threat to Apple, which was hard at work trying to get the Mac interface to work with A/UX, and Microsoft/IBM, trying to bring OS/2 and Presentation Manager from the realms of vapor, the two companies announced a joint venture in developing yet another windowing interface: Open Look (see “AT&T Opens The Windows,” MicroTimes issue #44, June 1988). Based on Xerox PARC graphical-interface technology, Open Look promises to deliver intuitive ease of use combined with the power of UNIX.
While the alliance offered obvious advantages to Sun and AT&T and promised to make life easier and more productive for their customers, it brought howls of outrage from hardware vendors like IBM, DEC and Hewlett-Packard, who license their own UNIX dialects from AT&T and perceived it as giving the two companies a virtual monopoly in the UNIX marketplace. The upshot of this was the formation of the Open Software Foundation, a group of hardware vendors who committed themselves and their resources to the development of an alternate UNIX standard, to be based on IBM’s AIX. There ensued something of a scramble for market- and mindshare among software vendors as OSF tried to convince them to develop applications for its environment rather than V.4. Subsequently, vendors loyal to the AT&T-Sun standard formed their own support organization, still unnamed but currently calling itself “The Archer Group.” (See News and Analysis, this issue.)
Sun, meanwhile, continues on an aggressive pursuit of excellence, spending 13% of its revenues on research and development, compared to the 8% industry average. “At any time,” says McNealy in the 1988 Annual Report, “Sun may have dozens of widely varying R&D projects under way, occupying well over a thousand engineers.”
The 33-year-old son of an American Motors vice president, Scott McNealy has been president, chairman and CEO at Sun for most of the company’s existence. His credentials include a degree in economics from Harvard and a Stanford MBA; he’s at once an effective executive in whose presence AT&T honchos tend to be positively deferential and a fun-loving golf and hockey enthusiast who remains Sun’s chief cheerleader.
An animated speaker, McNealy tends to make his points forcefully and with an utterly Sun-like sense of “Isn’t it obvious?” When we got to talking about the AT&T alliance, he reached for a tablet and scribbled a diagram of the current state of the UNIX market, with a glorious future for V.4 UNIX and trouble ahead for obscure variants like OSF and the Mach operating system of the NeXT machine.
Given Sun’s performance the past few years, there’s a pretty good chance he’s right.
Just to review some ancient history, how did you get involved with this gang of 8000 anyway?
The whole genesis of the thing basically started with a couple of engineering projects funded by universities and government grants. One was Andy Bechtolsheim’s work on the Stanford University Network and Sun workstation, and Bill Joy’s work for DARPA on the Berkeley version of UNIX. And then we had a couple of business school students that became close friends, myself and Vinod Khosla. Vinod was out looking to go do a startup; he met up with Andy and said, “Let’s go make a computer company out of this.”
I was, at the time, director of operations for Onyx, which was a company making UNIX minicomputers in the Valley. The two of them came over and visited me, and I said, “ Well, let’s go talk to this venture capitalist.” One thing led to another and pretty soon Vinod said, “Well, you’re going to quit your job, aren’t you?” I said, “I’m getting paid lots of money—why would I do that?” He said, “You can’t back out on me now!” So really, with no more than about 24 hours’ thought, I quit my job, joined them, and we went up and recruited Bill Joy from Berkeley. That was back in the February - March time frame of 1982. The rest is history, I guess.
Open architecture is important to Sun; how does that affect the way you compete in the marketplace?
Well, I came from the automobile industry, where if you bought a Ford, you didn’t have to buy a Ford for your next automobile. You really locked the customer in through customer satisfaction, and reliability and quality, and price/performance, and marketing, and styling and distribution channels — all those other things. You didn’t lock them in to the way the car operated, or the kind of fuel it used, or the roads it drove on.
If you go back to the ’82 - ’83 time frame, a lot of computer companies were complaining because their computers wouldn’t talk to other computers, and they were all confused that no one vendor could supply the whole range from bicycle to Boeing 747 in terms of their communication needs. And there’s an absolutely fundamental shift that’s going on in the computer business. If you look at the two fastest growing computer companies ever in the history of the world, it’s Compaq and Sun, both of whom are dedicated to building cloneable computers.
Doesn’t that mean that you have to stay a couple of steps ahead of your competition?
Just like Ford! Ford is a wildly successful company. And I hope to be that “unsuccessful” and face those kind of competitive pressures someday. I hope to be that big someday.
I'll tell you what—you can’t go very long in any industry, any place, anywhere, and not be a couple of steps ahead of your competition in price/performance, quality, reliability, distribution, marketing, all of those things. People are always looking for that magic little business where you don’t face competition and you can make wild, crazy profits. It doesn’t exist! There’s no free lunch!
I appreciate your asking that. That’s always the classic response to, “Well, so you’re building cloneable computers, aren’t you going to have to face competition?” Compaq has more than faced it, and their products are much more cloneable than ours; much more cloneable than an automobile. You can build a Compaq clone; in about six weeks you can be shipping. You can’t do that with an automobile.
Does that mean that one of the primary dimensions that you’re competing on is technology—staying at the forefront?
We’re spending 13-, 14-, even as much as 15% of our sales dollars on our R & D budget to build new, better, faster, smaller, more capable, state-of-the-art machines. That percentage is far above the industry average in terms of investment in new product and new technologies.
You said that you’d like to be as big as Ford someday.
Well, that’s already taken out of context. I just used it as an example. It’s not a goal. We have a saying around here: “Goals only limit you.”
It would be highly misrepresentative ofthe company to put a statement up that says “Sun wants to be this big,” or whatever. What Sun really wants to be is respected, successful, highly ethical, a fun place to work. The kind of place where you're proud to go to the outside world and meet some of your friends, and they ask you where you work, and when you say, “I work at Sun Microsystems,” they go, “Oh, wow. That’s neat.” That’s a more important goal here. I think a by-product of that is growth, profits and all the rest of it.
At the same time, the growth and the profits create their own problems in a way. It’s easy to hold onto that edge when you’re a small, little guerrilla company, but when you’re taking over Mountain View...
I don’t know how to deal with it! You can’t win with the world, right? You go be open, and you get a chance to lose, and people say, “Well, now you might lose to the clone manufacturers.” You grow, and now they go, “Oh gosh, now you’ve got growth problems.”
Well, I’ll tell you what. I’d rather be wondering where I’m going to get my next person as opposed to wondering who I’m going to lay off, or what project I’m going to cut because we’re not growing. So, you’ve got to pick and choose your problems. We happen to think that growth and being able to lose are much better problems than being totally locked out. If you look at a company like a Data General, there’s one built on proprietary architectures that doesn’t even get to lose. And they’re facing the other problem — not the problem of growth.
But take, for example, Apple. Everything is getting very chaotic, they’re hiring people faster than they can train them, employees are demoralized by the constant reorganizations, they’ve got executives feuding with each other... So, how do you propose to avoid falling into that pit, because it’s a very natural pit to fall into?
I think Sun has a foundation, has a tradition, has a history and has a management team that is much more fundamentally—MBAish, or process-oriented, or people-oriented, and sound-business judgment oriented, than just about any other classic Silicon Valley entrepreneur, young-kids-oriented kind of startup.
I mean, Sun hasn’t had that bad quarter, hasn’t crashed and hit the wall, hasn’t had massive turnover, hasn’t had that time when we absolutely had to regroup and refuel and reorganize. I don’t know why. You'll have to decide why those others have had those situations and why we haven't.
I like to think that Sun has a real dedication to sound fundamental management. It’s real boring. But we spend time to go do things like go one-on-one, have staff meetings, do planning, do reviews, chart where we’re going, have a budget process. You know, all of the things that are just fundamental Management 101: do reviews of employees, pay people for performance. I think there’s a very squeaky-clean focus on honest, fair, ethical behavior that keeps people from getting frustrated.
The worst thing a manager can do is get in the way of their people. And management here tries to stay out of the way of the people who are doing the real work. When I go into a one-on-one with somebody who works for me, I say, “What can I do to make you more effective?” As opposed to, “Will you do this? Will you do that?”, etc. It’s a different relationship, it’s a different focus, it’s a different management style. Why are companies successful? I think it’s because of the management culture and structure that they put in place.
A lot of people think that you’re kind of young to be running a billion dollar company.
I’m seriously young, not kind-of.
Well, the common wisdom is that the best CEOs have already screwed up two or three times, so they’ve gotten it out of their system. What do you do to compensate—or do you disagree with that?
I probably make more mistakes in a week than most CEOs make in three months. But-I’m certainly willing to acknowledge that I’ve made them and go back and try and fix them. I think that’s an attitude and style that exists in this company. It’s all right to make mistakes—don’t make them twice. Let people know what you’re doing. It’s a very participative management style here. I try and make as few decisions as possible. That way I’m less likely to make the wrong ones.
People often talk about consensus management. We don’t have consensus management here at Sun—it’s very line oriented. And the line organization makes the decision. And if they own all aspects of the decision, wherever that comes together in the organization, that’s who makes that decision—if it can’t be pushed down even further.
The one thing we are very much dedicated to is participative management. Everybody knows what decisions are going on—the whole world knows what we’re dealing with. There are more rumors, there is more written about, there is more activity in the press about what we’re doing, and they’re all looking for that magical, “We've figured out what Sun is doing!” They don’t understand they’re just hearing bits and pieces of a very open, participative process within the company. And yeah, every rumor I’ve ever heard about Sun has been accurate in the sense that, yeah, we've evaluated it. But they think that because we’ve evaluated it, it’s a direction we’re heading. And it’s totally misleading.
We used to worry that people would come in and see our email by logging into our network. Once somebody had an email conversation going on about “How are we going to protect our network?” We said, “It will confuse anybody so drastically if they ever read anything at all that goes on in our email”—just because — that isn’t necessarily a decision. There’s a lot of discussion, there’s a lot of participation. But then we go make the decision and away we go! And then we announce it. No hidden agenda. It’s just, we don’t know yet. When we say, “We don’t know,” we don’t really know. When we know, we want the whole world to know so we can tell everybody.
But that’s really a management style that was generated out of ignorance and inexperience when I took over. Three months earlier, I didn’t even know what a disk drive did. The last thing I was going to do was try and make a decision, so I tried to get everybody who knew in a room.
You’d be surprised at how basic logic solves most, if not all decisions. And then the rest of the decisions are lose/loses—there’s no hope. They don’t bring you the easy ones. They solve the easy ones and they only bring you the impossible lose/lose situations.
So you just decide whether you want to lose your right arm or your left leg, and just make some kind of fundamental judgment on which will hurt you less over the longer term. Those are the tough ones that I have to get involved in because nobody wants to be the one that says, “This is gone, or that’s gone.” Those are the ones I have to make. They’re not fun, but they’re aired out enough so people understand...in fact, they’re pleading with me to make a decision because we’ve just got to move on. We’ve got to get it over with and start growing a new arm back.
There seems to be a merging between the workstations and PCs. The 386i seems to be some sort of recognition of that. What do you think will happen in the next few years? Will you go to retail channels?
There has been, at Sun, a very clear statement for a couple of years now; people aren’t listening, because they don’t want to.
But we don’t talk about PCs or workstations anymore. Sun is a general purpose computer company. We are one of the top ten computer companies in the industry, in the United States, and we only refer to desktop computers, networks and servers. There’s no such thing as a class. There are desktop computers that are anywhere from a calculator to a high-end 3D, solids modeling superworkstation, or whatever you want to call it. But it’s $70,000 to $7 — that’s the range of desktop computers you can buy. There’s a blend all the way across of different price and performance and functionality points. So I can’t answer your question in the sense of PCs versus workstations.
Well, let me put it another way. You can buy “PCs” through traditional retail channels...
You can buy desktop computers through many different channels. They can be installed direct, they can be installed through OEMs, they can be installed through VARs, they can be installed through Businessland. You can actually mail order them. That is so misleading, and that is so...inflammatory!
It is just inflammatory to say that Sun is going into the PC business! That sends our stock price down! That’s why I’m so adamant about this! Because people think that it’s something strange or weird. Sun is doing desktop computers! Deal with it, world! We’ve been doing it for a long time, we’ll continue to do it! And we’ll take it through whatever channel is most cost-effective to take our technology through. So far it’s been direct and OEM, and we're expanding now into the VAR channel very aggressively and very successfully. And we think that’s going to be a very good channel.
We happen to have some products that go through the dealer channel. Those products have typically been software and board-level products. They’re selling through our TOPS organization. Now there are also Sun products going through the private-label channel; Fujitsu is a classic example of that. They’re buying hundreds of millions of dollars. So there are many, many different channels, and there will continue to be many different channels that will take our products to market.
So the two questions you are asking are, “Are you going to build lower cost desktop computers?” Absolutely! We’ve said that from Day One; we will continue to do that. We will continue to drive the price/performance down to lower price points. And we think we can build very competitive products.
At that point, are there other, more cost effective distribution channels to deliver it? And maybe there are some new ones that become more important. Go back 10 years ago, there was no such thing as a dealer channel. Ten years from now there may be a whole new way of delivering this product. We'll stay very flexible. Will our product end up in a ComputerLand or a Businessland, or something like that? Those conversations have been going on for about five years now. They will continue to go on, and everybody wants to be the first to predict when it happens, and it’s not useful.
Anybody who needs any of our products—they’re totally available. Anybody who needs to get ahold of our product can get ahold of it.
I was trying to highlight the fact that Sun is now competing more head-on with the Macs and the PCs, as opposed to a few years ago when there was a pretty clear distinction.
There never was in our minds. It made it more convenient for people to write about the industry. But when the buyer tried to decide what they wanted to put on their desktop, it was very confusing to them. Because they wanted the Compaq or the IBM price point, but they wanted the Sun functionality. Now they’re starting to see the marriage of the two.
The only reason I get so enthusiastic about this whole concept is that...I just read that Stewart Alsop was quoted as saying, “It’s going to cost Sun a billion dollars to get into the PC business.” I have no idea where he gets a billion dollars, nor what it means to get into the PC business, because I don’t know what the PC business is! If you ask an engineer who has a $50,000 dedicated Sun workstation on his desk, it’s a personal computer to that person! It’s theirs! They own it, they run it, they manage it. It’s their computer! It’s a computer and it’s personal. Personal to me implies a one-to-one relationship.
Are we going to get into the Intel 386, DOS business? We already are! Are we going to get into the $2000, 286 clone business? I can very comfortably say, “Why?” So you have to look at each little piece of the desktop pie, and say, “Is that the one you're going to go after?”
Can we talk about Open Look? It seems like from at least three directions, people are converging on the same thing: A friendly multiuser, multi-tasking environment that’s easy to deal with. There’s OS/2 out there with Presentation Manager...
There’s forty different dashboards out there and you can operate every one of them because fundamentally the concepts are similar enough and intuitive enough so that you know, “Well, that’s a radio, and you can turn it on.”
From a user’s standpoint, what is the advantage of Open Look over some of the other options, all of which have the disadvantage of being a little vaporish at this point?
The user doesn’t have any advantage until the product ships. From a user perspective, Lotus 1-2-3 written to Open Look, versus 1-2-3 written to Presentation Manager versus Excel written to the Mac—they’re all pretty much going to be six of one, half a dozen of another in terms of the training time. You sit in an automobile, it takes you a few seconds to get familiarized with the dashboard. It will take you some minimal amount of time to get familiar with the dashboard, if you will, of this new Lotus application or Excel application. Ifyou know how spreadsheets work, you'll be able to bomb your way around.
Now the one advantage you can look to is performance, and the features and intuitiveness of the environment. And Open Look, we believe, has more intuitiveness, in the sense that it has things like a tack pin, where you can pin things up as opposed to having them go away. So, you have a push pin that you can push into your little document—that’s very intuitive.
It has things that are very productivity oriented. If you want to scroll up in the thing, in a Mac window for instance, you have to move the mouse up to the top right hand corner and then scroll up. And then if you want to move down, you have to move the mouse all the way down, travel all the way down the window to the down button and scroll down—instead of having an elevator which you put the cursor on and then you move your mouse up and down, and it scrolls up and down as an elevator. So, you can go up and down without ever repositioning the cursor. That’s highly productive with respect to that one little task.
Now you do a bunch of those things really well, and you can get an environment that is more productive and easier to use.
Then you have tool kits and the tools that allow you to build an application faster, which means if you buy a computer that supports Open Look you can potentially get applications faster because the developers can build the applications faster. You’ll have one that runs in more than one environment. Presentation Manager only runs on Intel machines. As far as I know, it’s only gonna run on Intel machines. Open Look is going to run on every instruction set in the business. So you have an environment that’s much broader and portable. The Mac look and feel only runs on Macintoshes. So Open Look will be much more pervasive across many more computer companies’ environments.
So potentially any machine that runs UNIX can run Open Look.
V.4 will have Open Look as part of the package. So any machine that runs V.4 will run Open Look.
The OSF people are forever bitching and moaning that you and AT&T have this lovely sweetheart relationship....
Can I draw you a picture? (starts drawing diagram on tablet)
This is total garbage and the press has been....I’m kind of aggressive, so I apologize, but I think the press has not had an incentive to solve the confusion. And in fact they’ve enjoyed this story because its probably sold more ad copy than anything. But it’s very simple. The world pre-OSF and pre-Archer Group was this: From a UNIX-centric point of view, there was the Coca Cola Classic, which was System V from AT&T. It was the only one with a trademark. It was the UNIX, it was the original, it is the only, right?
There’s a whole bunch of what I call variants. There were thousands of them, right? And [drawing chart] let’s rank them in terms of volume: XENIX, Berkeley, Sun OS, blah, blah, blah. I don’t know what they all were, Ultrix, HPUX, A/UX.. you get way down here and you have AIX.
Which IBM now claims to be very proud of.
Right, and it’s a non-volume variant. You have to understand that! It was and is a non-volume variant! And then you get down here to probably the lowest volume—Mach, which has maybe six units shipping, thanks to Steve [Jobs]. So, now we have another variant! The press absolutely whiffed on that! He just introduced another G-D variant! We don’t need another variant! This is the situation we’re dealing with. All right, so this is pre-OSF.
And then we had what I call the deviants. Highest volume there was DOS, then Mac OS, then blah, blah, blah, right. You know, then VMS, blah, blah, blah.. then VM. And then down here you have OS/2. And you pick your favorite going-out-of-business deviant down here. Now that’s the world.
Really confusing isn’t it? It always has been, always will be. At least for the next few years until we get to the internal combustion-engine and everybody does it that way. But it’s been confusing.
Now AT&T tries to do the right thing. They said, let’s take the highest two volume-variants, Berkeley/Sun OS and XENIX, and let’s combine them into V.4. Great. We now have, basically, only low volume variants left. And that’s true.
There is no high-volume variant that is more than a few percent of the UNIX volume left in the variant category. And this merges everything to V.4. And what you saw with the Archer Group is that 70+% said, “We’re not commited to a variant. We’re not commited to a process. We’re not commited to an organization. We’re commited to shipping this product on our workstations, on our computers, on our everything.” Now you have yet another variant being offered [OSF]. Which I put down here absolutely dead last. It has no volume. No volume! I don’t care about statements of...marketing statements. It’s a mutant AIX. It’s a mutant variant. Because they will mangle it, because they don’t want to pass through AIX and give this low volume company here [IBM] an opportunity to have an advantage. So this will be a mutant AIX, a mutant variant. And it will be out someday in the mid 1990s. The war will be over, the game is done. It’s all over.
Now what people aren’t saying, and won’t say, because they think IBM is a powerhouse in UNIX, is that there is nearly no chance for any of these variants to become even a low-volume variant, much less a significant volume! If you’re a software developer, you’re going to develop to this one first because it’s the one true product. There are software developers’ kits out on X11-NeWS [Sun’s windowing environment]. There are software developers’ kits out today on Open Look. And you can go to Sun, or XENIX, or a System V machine today and have a very nice migration path to V.4. And there’s no environment out here [toward OSF]. °
I mean, why would you do Ultrix, because they’re going to move to OSF? Why would you do HPUX? — they’re going to move to OSF. Why would you do A/UX?— just read about it. Why would you do AIX?— because IBM isn’t serious about UNIX, and it’s going to be different from OSF anyhow—which is where they’re supposed to be moving.
And what is IBM really doing, given that they’re not at all paying attention to the OSF practical user interface, because they went out and did something different with NeXT, and said screw you? I mean, they got them all stuck to flypaper and then ran off. I mean, there’s nothing confusing about what I just said! Tell me what’s confusing. Because you have to be convinced!
The thing that’s confusing is—what’s in it for them?
What’s in it for who?
OSF. From your standpoint, why are they even bothering?
There are two reasons OSF was formed. The first one, which has been solved, was that AT&T wasn’t listening to its customers. And AT&T responded. And it probably took those companies all spending $13 1/2 million to get it done. But they formed the UNIX division.
And two is that the Archer Group was formed as a user organization to tell AT&T how and which ways they ought to deal with the licensing terms and conditions — how they ought to handle the technology, etc. That has absolutely solved all the problems H-P had. And I think if you asked H-P, if this had existed nine months ago, would they have wanted to pay $13 1/2 million to another organization that is not likely to generate a product and where most of the dollars are going to IBM’s bottom line to pay for AIX? I think they would say no. But now they’re trapped. This thing blew up in their face. It became bigger than life. And now they’re trapped.
Now, that was one of the problems, and it got solved. As far as Unisys, NCR, Sun and others are concerned, AT&T has absolutely...we thank OSF. Thank you very much, you were very useful! We’re certainly glad we didn’t have to pay to make that happen! But, you know, H-P, you’ve done the industry a very nice favor. Thank you, now V.4 can proceed and become the standard that it is.
Now, the other problem, and the reason OSF was formed, was that DEC, IBM, H-P and Apollo wanted to throw a big curve ball at Sun Microsystems. And that probably will not go away.
Can we talk about SPARC and where it falls in your overall strategy? You guys are probably at least as much a software company as a hardware company. And now you're a microprocessor company too?
This is another one where the press has got it all wrong! I can’t, I mean...
We're so glad we came...
This is why I like to spend the time. Because every time I turn around they say, “There’s Motorola pushing the 88000. Intel’s pushing the 386, and Sun’s microprocessor, SPARC.” Sun has never sold a SPARC chip in its life! Never will! It’s TI’s SPARC! It’s Cypress’s SPARC!: It’s Fujitsu’s SPARC! You never see it written up! Because they’re competing against TI. Motorola isn’t competing against Sun in the chip business. They compete against TI. But nobody will ever write that! They refuse to. We don’t sell chips!
I actually think that’s really pretty clear. As far as you going out and designing...
It’s never written that way. I can show you two million situations where its... I’ve never once seen, “TI’s competing against Motorola in the microprocessor business.” Not once!
We're going to do some things to change that. But we'll give you a little salt pellet in a rain cloud so you can all kind of gather ‘round and make it rain the right way here. But that’s our job. Because we don’t do SPARC chips. We do the Sun-4 computer based on the SPARC chip. Just like we do Sun-3 computers based on the Motorola chip. And Sun 386i computers based on the Intel chip.
So SPARC was more by way of establishing a standard?
SPARC was a way where we could buy chips from multiple vendors across a broad range of technologies with RISC as the architecture, as oppose to CISC [Complex Instruction Set Computing]. That’s all it was.
People are looking for some sort of magic, like we’re trying to create a standard. We just wanted to go buy RISC chips where we didn’t have to pay one company and count on one company to make that architecture fly. We have seven of the best silicon companies making our V.4 architecture wiggle fast: Intel, Motorola, Cypress, Fujitsu, TI, BIT and LSI. If those seven aren’t on the leading edge, then I'll be real surprised.
What’s the difference between the chips you’ll buy from different vendors?
Oh, BIT’s building an ECL (Emitter Coupled Logic) chip that will put out 40 - 50 MIPs. We’re buying custom CMOS from Cypress...
Will they all run the same instruction sets?
They are all identical instruction sets. Now they can add enhancements to it, but the basic core architecture is identical.
What does “scalable” refer to?
The fact that its a very simple architecture that can be implemented on gate arrays, custom CMOS, bi-CMOS, bi-polar ECL, gallium arsenide. You can scale it across from very low end computers to very high end computers, from very low clock rates to very high clock rates, from very low tech technology to very high tech technology like gallium arsenide. You can build laptops to supercomputers with the technology.
Do you anticipate a lot of other vendors building SPARC-based machines?
Well, they’ve already announced. I mean you seen how many people have already announced: Unisys and ICL, Arix, Solbourne, Prisma, AT&T, Sun, Seiko and Fujitsu. There’s going to be more SPARC computers than...I really believe that there’s more investment, on the silicon side, in SPARC, than any instruction set ever in the history of the world. There is more across a broader range of technologies, across a broader range of companies. There’s more software written on the SPARC RISC instruction set, and more UNIX expertise, than any RISC architecture anywhere, by far.
We have the best UNIX teams with AT&T and Sun and Unisys and others working on SPARC UNIX. And there’s more application software, by far, written to the Sun-4 with the SPARC microprocessor, in the RISC arena than any other RISC out there, by far. There’s higher volume, by far. Do you know we’re doing about a third of our revenue right now with the Sun-4? That’s after a year of having that thing out in the world. And there are more SPARC computers under development than any other RISC computer development architecture.
And it doesn’t matter who wins, because this is a winner! This is one of the winners! I mean, you’ve seen how many operating systems there are. Well, there are going to be a lot of instruction sets. But this is going to be one of the clear high-volume architectures and one I think you'll see both the classic low-end software developers who develop personal productivity software—certainly all the current mainstream UNIX application developers, and even in the high end when they’ve typically been writing to mainframes, I think you’ll see everybody porting to the SPARC, V.4, ABI—applications binary interface.
So, people just aren’t acknowledging... mean there’s nothing I’ve said that is really refutable. Because people are always challenging me, and if I don’t have a very strong conviction I'll retract the statement, but I just can’t retract any of those, because there’s nobody who’s got anything close to anything, all through that spectrum. And yet people, for some reason, are just, I don’t know, waiting for the big boom. You know, it just can’t be accurate. But this is the same story we’ve been giving for years, and it’s only gotten more accurate.
It seems that you’ve fallen a little bit out of favor with Wall Street—what’s your opinion of that?
Oh, I think that’s a further confirmation of our strategy.
In what way?
If Wall Street says they’re scared, then it’s got to be working. If Wall Street doesn’t understand it, then were doing something right.
What’s your opinion—I have to ask this—of the NeXT machine?
Let me go through it. This is kind of a rote answer.
Wrong microprocessor. Why would you do CISC as opposed to RISC? I think he regrets it. You notice he didn’t talk about performance at all. Second thing—eight megabytes. This is hardly what I'd call a low-end student desktop computer.
Thirdly, wrong operating system. He’s got the second-to-lowest volume variant on the list here. Ahead of only OSF in volume.
Which doesn’t exist yet.
Fourthly, he’s got a totally proprietary approach. You can’t go do an Apple unless you're first. You can’t come in last and expect to be proprietary.
Fifth, one of his great advantages is his optical disk, right? We could write a driver for that disk in a couple of weeks and have it through our R&D and QA facility in a month and beat him to market with it, if we wanted.
Sixth is—it’s not shipping until next summer.
Seventh is—$6500 is his 40% discount price, not his list price.
Eighth, his sleek styling is nothing more than a big, clunky, ugly cube. Let’s call it what it is. It’s a very nice paint job — for black. But it’s nothing more than a big, clunky box.
His wonderful software development environment that allows you to build applications faster will be necessary, because everybody first is going to write to V.4, and he’s going to be way behind, because the first application implementation is going to be done under V.4. And then they’ll decide to do #2 to somebody else — probably OS/2 or probably Mac OS. So he’s going to be fifth or sixth in line, and he’d better have a way of getting that application out fast, because we’re going to have been shipping the low end applications for many, many months before it ever shows up in his environment because of the volume.
Finally, I think he’s going to be awfully surprised at the level of competition that he’s facing, summer of ’89, in terms of what new machines are out there. I saw his product being compared against the 3/50, which was introduced about three years ago.
Why bother to make that comparison?
Well, he has magic dust! He has absolute magic dust!
And he uses it liberally!
How many people got to play with the machine? It was behind a glass wall. Do you really need to rock on with your bad self when you program? Is Bob Dylan really strategic to programming your computer?
And is it not better to have the DSP as a daughter board plug-in, so that it’s optional, as opposed to something you’ve got to pay for if you're not going to use it?
There are a whole bunch of things...where’s his floppy? Would you call up Canon [who did the disk drive] and ask them how much the removable media really costs—not what he’s charging for it? Because it costs 2x what he’s charging for it!
Would you talk to him about his mainframe architecture? Which, if you sit down with Bill Joy and ask him, “What does he mean by mainframe architecture?”—he means DMA. You should have heard the conversation between Andy [Bechtolsheim] and Bill. Andy came running up to Bill and he goes, “My God, he’s got mainframe architecture.” Bill says, “What are you talking about?” He says, “Well, he announced that he’s got mainframe-like architecture and all this other stuff.” Bill says, “Yeah, you know what it is, Andy?” He goes, “No.” He says, “Its D-M-A.” Andy says, “We’ve been doing that for years. Oh.. Okay, no problem.” He runs back into his lab...
I mean, it’s absolutely magical that the guy hasn’t shipped anything. Look at the press he gets—unbelievable!
The other thing I think is important to understand is—you have to look at the financial structure of NeXT. He owns 63%, Ross Perot owns 16%, a few founders each own 2%, some universities own 1%. That leaves very little left over. Their burn rate is massive; they’ve got 200 people. That’s a $3 M a month burn rate. Think about getting to next summer with prototypes and all the rest of it — the burn rate he’s talking about. He’s got a huge, underabsorbed manufacturing operation. He spent $100,000 on his logo, who knows how much on his announcement?
How’s he going to hold onto his people with stock options that have to be diluted and are such a small percentage of the company? How’s he going to hire and hang onto people? Do you really believe he has a different management style than he had at Apple?
Those are just a few of the questions, thoughts and issues that I would consider around that. Where’s his distribution channel?
The price does seem pretty high for a student—maybe the ones who buy the Mac IIs. But given what he’s got, I’m surprised he doesn’t go to the industry.
No distribution channels. Do you know how expensive it is to build one? We have been growing and building one faster than it’s ever been built before. Yet we had stock options that were worth something—to give to people, to get them to come. His stock options aren’t going to be worth anything because of the burn rate.
Do you realize how much cash he’s used, and he’s gotten to zero dollars in revenue? We used $4 1/2 million in our first two years. We were a $40 million company before we used up $4 million in cash. We used another 10 to get to $100 M, and we used another 20 to get to $250 M. We went public on about $50 M in investment. He’s going to use that much up getting to summer.
Think about it. How can you make any money and hire the people you need to take it to industry? You can’t do it. It is fundamentally a flawed business model in the way he’s done it. It’s too late. It’s DeLorean.
As long as we're on these kind of subjects... The Presentation Manager was just released. Will the 386i support it in future model? Do you think it’s important?
Right now, we’re putting all our efforts and energies into Open Look. In the same way—why did we do another operating system on top of UNIX? Because there was software written for DOS that was never going to run under UNIX. And so we had to go do DOS.
We believe we won’t have to do anything other than V.4 as an operating environment, because all software that is important, high-volume, profitable software will have as one of its platforms V.4. Same is true of Open Look.
We believe there will be very few, if any, applications over time that...I don’t care about the no-volume, nonprofitable companies that can’t afford to port to Open Look. If you can’t afford to port to Open Look then you're not worth having on Open Look. It’s kind of one of these self-fulfilling situations. We believe that all important software will run under Open Look. If that doesn’t become the case, then we would look at other operating environments. I’m not sure it would be Presentation Manager.
Part of the reason we believe that is that software developers who are, for instance, competing against Microsoft, don’t necessarily want to be based completely on a Microsoft windowing environment.
Especially since I know at least one person who says that Microsoft Windows on a 386 is what drove him to the Macintosh.
That’s the neat thing, I think, about the X11-NeWS/Open Look environment—that it’s a high performance environment built for the desktop architecture of the ’90s. It’s built with high performance 32 bit microprocessors in mind, portability in mind, user programmability in mind, and it took us a little longer than most people to get this stuff out, but it was built with a much more 1990s foundation and structure and strength around it. And that portability is something you can’t add on to it, and we just think it fundamentally has the support and the energy and the V.4 momentum behind it. So, we’re in pretty good shape.
What do you think you’ll be able to put on our desktop in five years for $5,000?
I don’t know. You’ll probably have somewhere between four and eight multiprocessors. Those will probably be 50 MIP chips. So, you’ll have between 200 and 500 MIPS on the desktop.
You'll be able to calculate those spreadsheets really fast.
That’s a really good point. There’s got to be a lot of software work between now and then. But then you’ll start to be able to do true voice I/O on the desktop, and you’ll have much higher speed networks, and you’ll be able to access the server world in a much better way. It will take a while.
You know, Sun could have far and away an order of magnitude easier user interface than the Mac, because it has so much more power and functionality and bits and all the rest of it. It just takes a while for that software to get written. And it will be a while for that to happen. So it will get there eventually.
A complete shift of gears, since you’re on record as endorsing Tom Campbell for Congress—what would you like to see happening in the larger political arena which would be good for Sun and good for the industry?
I think the basic issues we’re dealing with now—one is the cost of capital. These are not new statements, but that is a root cause of the reason why the Japanese have a monopoly on the DRAM business. And therefore, the Japanese have the ability to control Sun’s growth. I
think there’s fundamental problems with the tax structure in this country, where we tax savings, not consumption—that we ought to move to consumption taxes, not savings taxes. Which requires a major overhaul. I think we have a serious problem with capital gains, and that we ought to encourage capital gains differential.
I think we need to find a way to address countries which have CEOs of industry in their government—that are effective. In other words, MITI can tell Toshiba to only ship so many DRAMs, and Toshiba will say, “OK.” Can you imagine Ronald Reagan calling Scott McNealy and saying, “Only ship 2000 workstations next month”? Can you imagine that?
I can imagine what Scott McNealy might say.
Oh, you can! You can! But in Japan, they go, “Hai!” [yes]
You know, absolutely, we need to find ways to deal with, and to compete in that environment, because MITI has shown its ability and willingness to exercise monopoly power with respect to DRAM. And every toaster in the world will be shipped with DRAM in five years—your TVs, your automobiles, all of your telecommunications equipment. Everything will have DRAM in it. And they have monopoly control over that right now.
Other than that, everything’s in great shape!
What computer ads would you like to see in the future? Please comment below. If you enjoyed it, please share it with your friends and relatives. Thank you.