(FORTUNE Magazine) – Andy Grove didn't give it a second thought that summer afternoon in 1978, after a scrawny, callow computer programmer and his burly, bearded partner dropped by his office cubicle to shake hands. Just another scruffy wonk with big ideas and a squeaky voice. At least the other guy looked old enough to drive. What was the name of their little company? Microware?

Thus began one of the most unusual and lucrative business relationships the American economy has ever seen. Within a decade, Grove, now 59 and CEO of Intel, the world's largest semiconductor maker, and Bill Gates, the chairman and CEO of Microsoft who is now, at all of 40, America's richest man, would wrest leadership of the personal computer industry from the grasp of International Business Machines and Apple Computer. In the process they would throw open the marketplace to swarms of hardware and software developers, creating tens of billions of dollars in personal fortunes and shareholder value and changing ways people work and play. In the first half of the Nineties, the two lords of "Wintel" operated in tandem to create ever more powerful Intel microprocessor chips and Windows software, fortifying their position as kingpins of not only the PC business but also the entire computer industry. Today, in the age of the Internet, they're still at it, methodically plotting to harness networked computers to transform everything from telephony to television.

Grove and Gates may soar like twin condors in the digital stratosphere, but their relationship isn't exactly a heaven-made match. For one thing, each personifies the word "stubborn." Gates has been known to smash in the dashboard of a rental car after a prickly encounter with Grove, and to this day the pair occasionally squabble like an old married couple. They share other volatile traits too. Both are piercingly analytical thinkers who combine hands-on technical smarts with take-no-prisoners business savvy. Both absolutely hate to lose.

In other ways, the CEOs couldn't be more different. Baby-boomer Gates, a son of Seattle blue bloods, grew up in a privileged milieu of country clubs, summer cottages, and private schools; Grove, as a youth in his native Hungary, endured first the anti-Semitic atrocities of the Nazis and then the most oppressive years of Josef Stalin's domination of Eastern Europe before escaping to the West in 1957. Gates played a lot of late-night poker while attending Harvard in the mid-Seventies and dropped out long before graduating; Grove taught himself English and worked his way through college and graduate school in the U.S., eventually earning a Ph.D. from the University of California at Berkeley. Now that the frenetic Gates has gotten married, become a father, and hit middle age, he is beginning to plump up and seems a little world-weary; Grove, on the other hand, is full of energy and looks trim enough to host infomercials for the latest fitness contraption.

Gates and Grove aren't soul mates. But clearly they have built a fascinating partnership. On a golden Sunday afternoon this spring in Palo Alto, the two showed up precisely on time to meet with Fortune. Grove had spent the morning bike-riding with his wife; Gates had jetted in from a weekend visit to a friend on the East Coast. They posed patiently for photos, then sat down to explain the ins and outs of their alliance: to needle each other, reminisce a bit, finish each other's sentences, and peer into the future.


Their fates were hard-wired together in 1980, when IBM chose Intel chips and Microsoft operating system software as the key components for its crash effort to build a competitor to the wildly successful Apple II personal computer. At first, Gates and Grove didn't appreciate that they could one day become the dynamic duo of PCdom. Then IBM bungled its stewardship of the PC standard, and Intel and Microsoft grew stronger and began working more in sync.

GROVE: 1986 is when IBM began to lose it. For reasons of their own, they were reluctant to get involved with our 386 microprocessor. That's when Compaq got into the act. Then in 1990, Microsoft split with IBM and introduced Windows 3.0. As these things happened, instead of being two junior partners of a senior partner, we became equal players without that senior partner being present. If you look at this as a molecule, a lot of the bonds to that third atom in the molecule faded and were rebuilt between our two companies.

GATES: Yeah, Compaq's decision to come out with a 386 system before IBM is the big transition. Both our companies really encouraged Compaq to not just be the leader in portables, which is what they were at that point, but to be the performance leader too. After that, there was bit of a vacuum in PC leadership and both of our organizations recognized the need and opportunity to step in and fill it. But one key thing to know about the chronology of our relationship is that there's been more time spent on Intel/Microsoft collaboration in the last couple of years than in all the preceding decade put together.

GROVE: Let me explain specifically how Bill and I work together. The typical meeting between us theoretically occurs quarterly, which in reality means twice or three times a year. We alternate back and forth at their place or ours. Beforehand, lower-level people who work with each other day to day argue back and forth and set the agenda, and then Bill and I get together with them. These meetings typically go four or five hours.

GATES: The meetings between Andy and me are agenda-driven. Here's a little taxonomy: We're working in high-performance computing--big network server systems involving the Pentium Pro and Windows NT, etc. Then you've got the office desktop PC and where that's going. And then you've got the consumer and home PC. Beyond that, we have this whole non-PC space [i.e., pocket organizers, interactive video servers, TV set-top boxes, pen computers]. I'd say probably 30% or 40% of the things we talk about are in that non-PC category...

GROVE: Where we've flunked everything we tried to do.

GATES: Our optimism springs eternal...In PDAs [personal digital assistants--industry jargon for pocket-size electronic organizers], for example, we both did a lot of work that turned out to be premature. There was a chip for a PDA that didn't get designed quite on time. And we had some software that ran on it that wasn't on time.

GROVE: But the main reason this is escalating now is because PCs are getting more complicated, and the dividing line between our activities is blurrier.

As computing broke out of the character-based mode and into graphics, video, audio, networks, and stuff like that, each of these technologies started its own nucleus of collaborations between us. That's been the story of the Nineties.

GATES: This is a very dynamic marketplace. I mean, we're not simply deciding things for the industry. We're helping all the pieces come together. There's a few industry-wide initiatives, like "plug and play" [a technology that makes it easier to hook peripheral devices like modems, scanners, or printers to a PC], where the fact that we both have big R&D budgets and sales volume is very good for us. So we put together an initiative. But we don't determine the variety of PC configurations that people offer or which are going to succeed.


It doesn't take the intellectual candlepower of a Gates or a Grove to recognize that the Internet is more than just the latest fad. Nearly everyone agrees it's the biggest thing to hit computing since the PC. That's why in the past six months Microsoft has totally rebuilt its strategy around the goal of infusing Internet compatibility into practically every product it makes. And that's why Andy Grove is more obsessed than ever with the possibilities of the PC as a communications device.

Gates's and Grove's enthusiasm reflects their conviction that there are technological limits to the Internet--limits that can best be overcome by Intel chips and Microsoft software. Gates has harsh words for the notion of a $500 Internet terminal and alternative, non-Microsoft operating system software--concepts championed by software's other billionaire CEO, Larry Ellison of Oracle Corp.

Grove seems more open-minded to the idea of building devices other than PCs to plug into the Internet. Bottom line, however, both are betting they'll be the ones to deliver the goods for the Internet to realize its promise.

GROVE: The Internet is like a 20-foot tidal wave coming, and we are in kayaks. It's been coming across the Pacific for thousands of miles and gaining momentum, and it's going to lift you and drop you. We're just a step away from the point when every computer is connected to every other computer, at least in the U.S., Japan, and Europe. It affects everybody--the computer industry, telecommunications, the media, chipmakers, and the software world. Some are more aware of this than others.

As exciting as that is, though, there's one big problem--telecommunications bandwidth.

GATES: Bandwidth bottlenecks. No question, that's the biggest obstacle to where we'd like to take the PC.

GROVE: The problem is, PC users are accustomed to the quick responsiveness of a hard-disk drive, and even the fastest internal corporate networks don't transfer data to your PC nearly that fast. We all know how slow they can seem, especially with graphics and video data.

GATES: Low-cost interactive bandwidth--not just to businesses but also to homes--is going to come very slowly. Sure, if you look 20 years out, you're going to have a lot. But in the next five years, the percentage of homes that will have better than plain old telephone lines won't go beyond 20% to 30%, even in the U.S.

GROVE: I don't even think that's going to be met...

GATES: That's the optimists' number. There's all sorts of hurdles between here and there.

GROVE: On the other hand, the capabilities of a computer will continue to race ahead, even as bandwidth growth is comparatively sluggish. What we have to do is use that processor power and clever software to create the illusion of lots of bandwidth.

GATES: The key question for Andy and me is: As the world goes digital, and bandwidth is limited, will faster and faster microprocessors be valuable and will great software be valuable? There are actually people out there--this George Gilder guy and some others--who seem to think that as you get more bandwidth, somehow you don't need faster processors, or you need less software to help you find and manage things. That's just not the way it is.

GROVE: Here's an example of what we think could happen. Digital satellite TV has blown onto the scene in the last couple of years. I'm painting speculative pictures, but I'll bet there are people figuring out right now how to put together a living-room computer, plus a satellite dish, plus a big hard-disk drive. The computer can watch your Internet usage and, in the middle of the night, use the satellite dish to download and freshen images from your top 200 Websites or whatever. It would stockpile copies of those Web pages on your hard disk.

Once you have that, it's the de facto equivalent of the fastest connectivity any corporation has right now. The beauty of it is that what I just described is the equivalent of a high-performance PC plus some clever software and a very big disk. Technically it can all be done today, which means it's going to take five years to do it broadly.

Which brings us to the next question: What about Larry Ellison and his $500 magic box? The so-called network computer or Internet terminal. I've been asked about it so much I blank out on it. You know, after the 99th time it gets a little old.

GATES: You're getting better at your answer, though.

GROVE: Continuous improvement. [Laughter]

GATES: After I answered the same question about 20 times, I decided to write an essay and put it on the Internet. It talks about if you want to design a machine and all it did was run the latest Internet browsing software, what features would it have to have?

The basic trade-off is if you have high-speed network connection, you can make do with the cheap, diskless PC. But if you have low-speed connection, like most people, and want to do any real computing, you still need a full-blown PC. Another thing people forget is that the most expensive part of a network computer is the connectivity cost. Even if the hardware is free, the effective cost of paying $30 a month over time is more than buying a state-of-the-art PC today. So anybody who believes that this $500 network computer is the answer for someone who can't afford a PC has got a little bit of a problem.

GROVE: Excuse me. The reason it is so hard for my colleague here to talk about these Internet terminals is that the whole purpose of their effort is to build devices that are disabled from running any conventional PC software.

GATES: That's their purpose.

GROVE: No disk, no DOS.

GATES: I don't disagree that there are a couple of design points that might make sense. A TV/Internet browser [a set-top box comprising a cable TV decoder, a modem, and software for surfing the Internet] could be interesting. A portable, put-it-in-your-pocket device could be very interesting.

GROVE: It would be possible to fashion lots of different appliances. I think appliance is a good term, because these will be single-purpose devices. One could be simply an Internet-browser terminal. Another could provide something like telephony over the Internet. There will probably be a large number of devices developed and experimented with, but eventually the market will settle down to a small number of useful functions.

There's one other element. When you assess how good these devices are, they invariably will be compared with how PCs perform on the same networks. If the PC improves as it always has, it remains a moving target the new devices will never catch.

GATES: The greatest surprise to me of the whole PC market in the last five years has been that as you get innovation, demand for the machine that is just a cheaper version of last year's cool machine has been so small. Now, that's not to say the market will necessarily be that way in the future. But, you know, there was a time when if you had a nice 386 PC with two megabytes of memory, you were cool. Today you couldn't begin to sell something like that. I mean, last year Packard Bell offered an $800 box with a 486 DX4-100 processor. It's not such a bad machine. And it didn't sell, even at that price.

So if the network computer falls between, say, a browser connected to a TV and a typical PC, I really wonder if it'll get critical mass. We'll certainly experiment. We're doing subsets of Windows that people can try on these things, just like Intel's doing chips for these things. But the top priority for us, just like for Intel, is to...

GROVE:...give a decent experience on the low-end PC and a rich and growing experience on the high end.


What will the PC of the future be like? Grove and Gates are intent on making the machine even more versatile--by providing better video, graphics, sound, and ease of use.

Gates envisions a PC for the home that is a cross between a computer and a TV. Part of a new class of Windows computers he calls Simply Interactive PCs, or SIPCs (Gates has no knack for coining monikers), it would combine attributes of a PC, a videogame machine, a VCR, and a cable TV control box.

Grove, too, believes the PC can subsume the television. He's also still enchanted by the idea of videophones, and wants to push Intel's ProShare videoconferencing technology into PCs in the home. So what keeps these guys awake at night?

GROVE: I don't think what people used to call interactive TV is ever going to materialize. Your interactive TV will be a well-connected PC.

GATES: That wasn't so clear three years ago, but that's what will happen.

GROVE: Why? Content of every kind is going to be created and stored digitally.We are close to that already. Within a few years you'll have to do something weird to get an analog TV signal. We'll be viewing and listening to digital content through devices that use a microprocessor and a combination of network connectivity and information stored on a disk.

Well, what we just described is a PC; an evolutionary, high-performance PC with a big monitor, big hard disk, and the best connectivity you can afford in your neighborhood. Maybe five years from now cable modems will be available--or satellite dishes, or maybe all you'll have is high-performance processors doing the same stuff on a regular telephone line.

Here's a case in point: This winter you're going to see PCs put out by many manufacturers that will allow two-way videoconferencing over regular phone lines right out of the box. People wonder why didn't we do this three years ago? Well, guess what? 486s can't do that. 100-MHz Pentiums can't do that. But a 166-MHz Pentium Pro PC can deliver a better live video image over ordinary telephone lines than your old 100-MHz Pentium computer could over expensive, high-speed ISDN lines.

Microprocessors will get even more powerful. That's our business. Next you'll have laser-disk-quality video built into every PC, and when that happens, that's where entertainment content is going to go. By the time anybody makes television interactive, the TV in your den is going to be supplanted by what Bill calls Simply Interactive PCs, which will look like a TV but will have all these other digital attributes.

GATES: There are other breakthroughs that'll come gradually: the ability to talk to PCs; the ability of the computer to watch and respond to you sitting there, to gestures, and so on. Changing the way people interact with the PC is profound. If it can really watch you and listen to you and read for you, it's a very big deal.

GROVE: That doesn't keep you up at night, though.

GATES: No, but if somebody else were to do breakthroughs like that a lot better than Microsoft, well, that keeps me up. We have to watch for technologies that have incredible potential and make sure we're doing the best work in them. Making sure we're a leader in the Internet is the technology we're really driven to now. That's not just a worry for the night but for the day also. What worries you?

GROVE: The biggest problem we face is growth. Things come in and out of fashion, and who knows why? I have a deep-seated conviction that our business has some of the characteristics of the fashion industry. You always have to come up with something exciting and new to stay on top.

I'm not talking about growth in the next quarter; I'm talking about whether we are expending enough energy to develop stuff that creates this feeling that you have to have it. What really worries me, then, is this: Let's not let this industry get boring.

So far, it isn't. The Internet has created new excitement and revealed unrealized potential, and some unexpected new competition too. It has injected new energy almost to the point of too much, but that's what we need. Every year we need to infect another five million people with that excitement.


The Gates and Grove alliance is not without friction. Both are high-decibel table pounders who wield sarcasm like a samurai sword. Their companies mirror their aggressive and sometimes intransigent personalities. Gates and Grove say occasional blowups are just part of the chemistry that makes the relationship work. Call it creative tension, or perhaps a form of therapy. Whatever it is, it hasn't slowed the companies down much.

GATES: It must have been in 1978 when Paul [Allen] and I first came to see you. It was a very big deal for us. I mean, Microsoft was just 11 people then, and Intel was 10,000, so there was quite a disparity.

We ended up doing a few development contracts for Intel in the late Seventies and early Eighties, even before the IBM PC. For the first few years, you and I hardly spent any time together. But at one point, I remember our Intel liaison didn't feel Microsoft was being, shall we say, responsive on a certain project. So I came to your house for a dinner. Our contact was rolling you out as the big gun to put me in my place.

GROVE: It was not a pleasant evening. I remember the caterer peeked into the dining room to see what all the ruckus was about. I was the only one who finished my salmon.

GATES: Yeah, we definitely had different views.

GROVE: After that, other people in our companies continued to do business, but the two of us didn't speak for quite some time.

GATES: A couple of years went by, and we had another dinner that kicked off a period where we met on a very regular basis.

GROVE: There was a meeting at a restaurant in Palo Alto in '83-ish when I made a proposition to you. Remember?

GATES: Yeah, we didn't really follow up on it, but there was this notion that Intel was spending a lot on software, and wasn't there some clever approach where we could help each other? Nothing concrete came out of it. Looking back, though, Andy's proposition was kind of ironic because Intel was part owned by IBM at the time. [IBM once owned 20% of Intel.] A few years later we proposed that IBM actually buy 30% of Microsoft. It was a real turning point when IBM said no. We thought, "Hmmm, I wonder why they're saying no?"

GROVE: When was that?

GATES: This was '86. When did IBM sell their stake in Intel?

GROVE: They sold the last big chunk of it in '87. They bought it in '82, so in '86 they would have still owned some part of us. Anyway, they said no because they already had their own [PC software plans], right?

GATES: Yeah, they always had efforts to kick us--er, well, to be completely on their own. But that was when it became clear they were pretty sure they were going to go off on their own.

There was another famous meeting in the mid-Eighties--I'm trying to remember where it was--when we got involved in Intel's effort to build a RISC chip [a radically simpler, faster genre of microprocessor].

GROVE: The San Francisco airport one?

GATES: That's the one. [Compaq CEO] Rod Canion was there. And [Intel chairman and co-founder] Gordon Moore was there. You could say it was the opposite of that dinner at your house. This time I was giving Intel a hard time about, "Hey, you'd better pay attention to these RISC guys."

GROVE: This was sometime in 1989, and Canion was acting as the intermediary between us. To me, his pressure was as significant as yours. That's when we started parallel development--you know, the leapfrogging development of successive generations of microprocessors. We doubled our R&D spending rate on microprocessors after that meeting...

GATES: Also, you got a lot more sophisticated about measuring and improving chip performance.

GROVE: This happens a lot, you know. One of us goes out and hears things that the other either doesn't hear or doesn't want to hear. We do a good job amplifying these things to each other.

GATES: Our latest disagreement was last year, over NSP [a pet initiative of Grove's, called native signal processing, to write software to help new PCs wring better multimedia performance out of Intel's microprocessors]. Your guys developed software to give PCs more multimedia capabilities without having to add extra hardware. It was a great idea, but the rub was that your software couldn't work with Windows 95. You had your own method for handling 3-D graphics and audio, and so did we.

GROVE: These differences occur because our business priorities are fundamentally different. We sell a new chip with a new computer; we don't sell many replacement chips or upgrades. Bill sees a big installed base to take care of.

Last year Microsoft was involved with the biggest thing that they'd ever done--the Windows 95 launch. They were obsessed with it, and that's probably understating it. Chip performance issues were of secondary importance to Microsoft. It's not that we had a different notion of the goodness of enhanced multimedia performance. We just had different senses of urgency.

I admit we were dumb enough not to understand that the software we developed was actually contrary to some of the features of Windows 95. And hence came all that crap.

GATES: Yeah. Last summer you and I had dinner in San Jose when--for the first time, really--I was able to articulate the problem the way you just did. I was able to say, "Look, we don't disagree with your guys. We think they're smart. But this stuff does not work well with Windows 95." So we said, "Given that we agree on goals, can we share in the development of these things?" Intel deserves a lot of credit for stepping back.

GROVE: We didn't have much of a choice. We basically caved.

GATES: No, no, you didn't cave. Come on. If Intel had shipped that NSP stuff, you wouldn't have done yourselves any favors.

GROVE: We caved. Introducing a Windows-based software initiative that Microsoft doesn't support...well, life is too short for that.


Gates and Grove are clearly proud of their companies and their own accomplishments. You'd never catch either saying, "Aw, shucks, it isn't that big of a deal."

Still, they aren't the kind to sit back and gloat or even reflect much about their success. They're too busy mapping out new goals. Nor is either CEO under the illusion that he could have pulled it off without the other's help. In fact, as time goes on, both men have developed a deeper appreciation--a sense of awe almost--for the power of their unique partnership, both financial and in the broader context of the computer industry. So what is it that makes them and their companies so special? Why is it that they work together so well? What do they really think of each other?

GROVE: I don't want to speak for Bill on this--we haven't rehearsed this, I swear--but we both have gone through watching minicomputer and mainframe people dismiss new computing phenomena, and look what happened to them. I don't think it's in our blood to be dismissive of new things.

GATES: We've seen many, many examples of companies who weren't open-minded. We saw companies like Lotus or WordPerfect ignore the graphical interface. They could have afforded to put five people on the thing--even if their CEO thought it was stupid.

Every day we get up and say, "Hey, what should our company be more open-minded to?" I mean that is an absolute mindset. Why over the years have we been having all these meetings about PDAs and set-top boxes? Because we can afford to...

GROVE: Also, because we can't afford not to.

GATES: Another thing that makes us unique is our mutual dependence. The other day someone asked me, "Can't Microsoft work with people so that they can be successful too?" That night I looked at a chart comparing Intel's valuation with our valuation over the years. Although they vary somewhat, in both cases they went from a relatively small number to a relatively gigantic number. And I thought, "When have there ever been two companies with that kind of dependency both rising to that kind of success?" Even though we can tell you about all these disagreements and sarcastic meetings, we haven't really gotten in each other's way all that much. To me, that's really amazing.

GROVE: It's pretty hard to create new stuff. You have to vector two major organizations' energy in one direction before you have a chance. And when we're not lined up for whatever reason, new things take a long time to happen. I mean forever.

There's a book coming out this year called Co-opetition. The authors apply game theory to business, and their thesis is the concept of complementors--companies whose business is not driven solely by competition. Microsoft and Intel are defining examples. The point is that when companies look at their interests in a too narrowly defined fashion, they don't win. Through all these processes between us, we enlarge our respective viewpoints to include the other person's viewpoint. It takes our obnoxiousness or your sarcasm to make us understand that there's another interest at stake. When that gets internalized on both sides, then the complementarity of the process works.

GATES: I still say we're unique. If you take the top 30 or so companies in terms of market capitalization, and ask which of these are complementary to each other, it's hard to find some. Coke and McDonald's have done some wonderful marketing things together and happen to like each other, but it's not core to the businesses of either one.

GROVE: Right. We are complementary in an arcane and detailed technological fashion. I can't even think of another example. Possibly in the 1800s the railroads wouldn't have developed so quickly without the telegraph or cattle ranching, I suppose. But then, maybe it will be hard to remember our relationship in 100 years too. I hope not.

GATES: But I think we should make it clear that there's no exclusive tie going in either direction. If somebody walked into Microsoft tomorrow and said they had a microprocessor that's cheaper and faster...

GROVE: Well, you've done it. You've tried it. You didn't succeed, either...

GATES: And vice versa. Intel aids and abets non-Microsoft operating systems.

GROVE: Still, we have our biggest successes when we apply our muscle and their muscle to the same problem.

GATES: It's fun to hear Intel's plans, because when they decide to do a next-generation processor, the execution is amazing. There's so much behind it in terms of capital and design and testing and the like.

That's not to say we admire everything about Intel. Whenever Intel comes in and tries to tell us something about software, it sometimes...well, we kind of shake our heads. They have the confidence and aggressiveness and other characteristics that come from their world-class activities in the chip-related businesses. But sometimes when they apply it to the software industry, well, you know, they'll miss something there.

But we always have to have respect for what the other company does. And there are many examples where we haven't appreciated what they understand, and vice versa.

GROVE: What I'm most impressed with about Microsoft is that they are superb tacticians. They zigzag very, very well.

And you know, if they are wrong, they can be very, very pragmatic. What they are doing in the Internet field is phenomenal. I don't think any other large company could have turned as profoundly and as broadly and--and it's not just bullshit. There are a lot of companies that just put slogans up in the air. But with these guys, when some situation turns up, it's like antibodies approaching the problem from all different levels of the company very, very fast.

We have emulated it a little bit, but they are far more skilled. A big part of it is the result of E-mail-enabled management, if you wish. They use E-mail very, very effectively to manage this rapid response. I've always been and still am envious of that skill.

On the flip side, they are terribly defensive. Bill is, and they are. Unless it's a particularly benign set of circumstances, the slightest difference of opinion--not always, but often--will create very defensive emotional responses out of them that kind of shut down productive discussions for a while. After doing this together for 15 years, we've learned that it will blow over; just stay cool. Let them sputter away, and sooner or later they'll listen.

But I have to admit that lot of our people at Intel are like that themselves. So when you take their defensiveness and our obnoxious arrogance and mix them all up, it can be a pretty potent mixture.

GATES: No black eyes, though.