You’re Going to Do What?!
JANUARY 16, 1986, 10:00 A.M.
Compaq’s strategy team gathers in the boardroom of our new headquarters in northwest Houston to discuss a critical issue. One of our engineers has picked up on signals from Intel that seem to present us with an incredible opportunity that’s also incredibly risky.
Hugh Barnes, one of Compaq’s top engineers who works closely with Intel, opens the meeting. “I believe IBM isn’t actively pursuing the development of a 386-based PC right now. Intel hasn’t said so directly, but they seem to be looking solely at us to help make it compatible with the 286 chip.”
I ask, “What’s IBM up to? Why aren’t they helping?”
Barnes answers, “They must’ve decided they aren’t ready to move up to the next processor yet. And based on how much work it’ll take to get a 386 PC to market, I think it’ll be at least another year before they could be ready.”
“Wow! That’d give us a pretty big lead. But what happens when IBM introduces their 386 PC and it’s different from ours? A lot of companies have gone down because they moved out in front of IBM and then got run over.”
Barnes replies, “Well, that’s the big question. There certainly is risk, but we feel it’s an opportunity we should take a serious look at.”
I ask, “So is the 386 really compatible with the 286?”
“Not yet. Intel is trying to make it compatible, but they’ll definitely need our help to get there. The last chip they sent us still has lots of problems they weren’t expecting.”
“How confident are you that together we can achieve 100 percent compatibility? Without that, the odds of us succeeding would be extremely low.”
Barnes replies, “I believe we can get there. But it could take a while if we don’t find all the bugs at this stage. Each time they have to cycle a new chip, it adds months to our schedule.”
I turn to Stimac and ask, “What about the software? Do we know of any killer issues that would keep us from running all the existing software?”
Stimac replies, “We don’t know of any, but there could very well be some. The 386 runs at a higher clock speed, so there’ll definitely be issues. I think we can do it, but we need to look at this very thoroughly.”
Then I turn to Swavely. “Mike, how do you think the market would react to the first 386 PC coming from Compaq instead of IBM?”
“I’m not sure. We need to do some research before we commit to this. We can talk to some customers we’re close to and get their reaction. But we need to be very careful.”
I nod in agreement. “OK. This is a great opportunity, but we have to get more information and think it through. Also, we need to find out if Intel is working as closely with any other company. I’ll call Andy Grove [president of Intel] and see if he’ll commit to supporting us if we agree to launch ahead of IBM.”
As the meeting breaks up, I’m already thinking about the possibilities this opportunity could create. My intuition is that we should do it, but my experience tells me to be careful. The result of getting it wrong would be disastrous.
We weren’t strangers to taking risks. Early on, we developed a process for making critical decisions we simply called “The Process,” but there was nothing simple about it. “The Process” evolved from a style I had developed as a manager at TI, which involved leading a small, carefully selected group of experts along a path toward consensus. The leader used the team and the consensus process to prepare him or her to make a better decision. “The Process” differed from typical consensus management because the leader retained the responsibility for making the final decision. The leader was responsible for keeping the group from getting bogged down in extraneous issues or dead-end paths. The leader also decided how long the meeting would last, and how many meetings would be held, before making the decision. Finally, accountability for the results remained with the leader.
The executive team received regular updates on the 386 at my staff meetings, which were normally held every Tuesday morning. There were also impromptu meetings called as issues arose that needed immediate attention. The excitement over what this new product could mean for our business continued to be tempered by the risks that could not be eliminated.
MARCH 12, 1986, 10:00 A.M.
Compaq’s product strategy team gathers in my conference room to make a decision on the 386. I open the meeting and nod to Swavely.
“We’ve looked at all the issues and believe we should do it,” he says. “But if we’re going to do it, we need to commit now and get going because there are lots of things we’ll have to line up before we can launch.”
I ask, “What’ve you learned about customers’ willingness to buy a 386 PC before IBM blesses it?”
He smiles. “There seems to be an insatiable demand for more performance. If it can significantly increase the speed of existing software like CAD packages, people will buy it. There’s still the issue of IBM changing the game at a later point, but as you keep telling the press, if their 386 PC doesn’t run all the current software it’ll be viewed as a disadvantage.”
I look at Barnes and ask, “How is Intel responding to our feedback about the compatibility problems?”
“They’ve really started paying attention. Early in the year, they wouldn’t let us talk directly to their main processor architect, John Crawford. We had to write it all down and they’d submit it to him. Then something changed. Last month we were allowed to set up a meeting with Crawford, and now things are really moving.”
“That’s great. Maybe my call to Andy Grove did some good. Andy said they’d work closely with us to get the 386 to market. He didn’t actually say we’d be first, but the implication was clear. I think he’s starting to understand how important we are to getting his 386 chip to be compatible with the 286.
“Are there any other critical issues to consider?”
Swavely answers. “Yes. Chip availability. They made so many changes to the chip during this pass, Intel isn’t willing to start enough wafers through their process to get us the number of chips we need to do the launch.”
I frown. Everything seemed to be lining up, but this is a potential killer. “Is there anything we can do about that?”
“We can buy some ‘risk’ wafers,” Barnes says. “We’d pay them to start additional wafers through their fabrication facility so that if the last version of the 386 chip they run actually works, we’d have enough chips to launch. In other words, we take the risk on the chip not working, not Intel.”
“They don’t sound very confident. Should we do it?”
“Absolutely. The money isn’t really that much compared to what we’ll be investing in everything else.”
I’ve heard enough. “I don’t want to take any chance on missing this opportunity. Let’s tell them we’re committed to doing it, but we want their commitment to work with us to make sure the chip is fully compatible. Even if it means some amount of delay in production.”
With that decision, we dove headlong into the swirling river of fate. We were trying to achieve something no other company in an “IBM-compatible” universe had ever achieved: We were going to try to take the technology leadership position away from IBM by introducing a major advance in processor architecture ahead of them. It’s hard to fully grasp just how much we were betting when we made this move. We had built an incredible reputation based on our consistent delivery of quality, performance, and business success. And we were about to risk losing it all if this move failed.
When I called Ben Rosen, our board chairman, to tell him what we had decided to do, he responded, “You’re going to do what?” He was shocked at first, but after I filled him in on the details, he shared our excitement.
It was not as though we didn’t have a choice. We could have easily told Intel no and waited for IBM to deliver the first 386 PC. That way we would have “played it safe” and made sure our product was fully compatible with IBM’s before we took it to market. At least we thought that would be safe.
The key piece of information missing was the real reason IBM decided to delay introduction of its 386 PC. As we later found out, IBM was working on its own completely proprietary PS/2, the PC industry’s equivalent of a “Death Star” that was intended to cripple or destroy every PC-compatible company on the planet. And IBM was planning on launching the new product line with a 286 processor.
If we had known about this then, it would certainly have affected our thinking, but it’s not clear what the effect would have been on our decision. We might have felt even stronger about pushing ahead with a fully compatible 386 PC in order to build as strong a market position and reputation as possible before the “Death Star” became operational. But we also might have decided it was better to wait and see just what IBM would do, because the risk of IBM’s new PS/2 making our 386 product obsolete was too high. Fortunately, we took the perceived “risky” path, which turned out to be the only one that could lead to long-term success.
Once the decision was made to move forward with the 386 PC introduction, every part of Compaq began to work toward making it a success. The implication of delivering the first 386 PC was we had to prepare the industry for its arrival. Whereas IBM had always been the one to run advertising and publicity campaigns to educate the market about the advantages of a new processor, now we would have to step up and take that role.
We became the focal point between Intel and Microsoft for making sure all necessary hardware and software components were ready and fully tested when needed. Leaving nothing to chance, we began to secretly work with all the important application software companies to make sure their programs worked and took full advantage of the 386 performance. It seemed to me that every part of the company was being stretched to its limit, but it was a role we had been preparing for since the beginning.
It seemed to me that every part of the company was being stretched to its limit.
Planning for the announcement event was stepped up to a new level. I wanted to anticipate all the important questions that might be asked at the 386 announcement and thoroughly answer each one during the presentation, before they were asked. I expected there would be a high degree of skepticism by the media and analysts, so it was important to have facts, graphs, and personal testimonies integrated into the presentation. Although most of our product announcements had required less than thirty minutes, we were going to have to work hard to keep this one under a full hour.
One question particularly concerned me: “What happens when IBM introduces a 386 PC that is different from yours?” I wanted to be able to answer that one so convincingly no one would be left with any doubt. We decided to have key industry leaders speak during the presentation, strongly supporting our new product’s importance and usefulness. This was intended to project the image that Compaq’s 386 PC would succeed no matter what IBM did.
The industry leaders we recruited were truly the “Who’s Who” of the PC industry: Bill Gates, chairman of Microsoft; Gordon Moore, chairman of Intel; Ed Esber, CEO of Ashton-Tate; and Jim Manzi, CEO of Lotus Development, creator of Lotus 1-2-3, the most successful spreadsheet application at the time.
By early September, all the stars were beginning to line up. The effort to get them there had been nothing short of monumental, but no one at Compaq was complaining. We were standing on the threshold of the future.
SEPTEMBER 9, 1986, 11:00 A.M.
We attract several hundred PC industry luminaries, media personnel, and analysts to the Palladium, New York City’s popular nightclub, to announce our most important new product up to this point. They all know the product is a 386-based PC, but they’re still excited because Compaq’s product announcements have evolved into events that are more like parties.
The presentation opens with Rosen, who puts the significance of today’s introduction in a broad context. During his remarks, he makes a bold assertion. “Today’s launch is far more strategic and far more important than any prior industry announcement.”
He explains why our launch is important to the personal computer industry, the software industry, end users, new users, and dealers. “This product will be the first to have the power to handle the revolutionary new software coming . . . that’ll make the personal computer so easy to use it will become as ubiquitous on the desk as the telephone.” Although many in the audience view this as hyperbole, it’ll prove to be accurate in the coming years.
Rosen then introduces a peppy multimedia show set to “Headed for the Future” by Neil Diamond. The show draws on analogies to similar leaps forward in technology in other industries, including automobiles, telecommunications, and medicine. As the multimedia show ends, an oversized mock Deskpro 386 descends to the stage, literally filling the entire space.
I step to the podium. “Ladies and gentlemen, introducing the Compaq Deskpro 386 personal computer. Today, Compaq Computer Corporation is stepping to the forefront of the personal computer industry. The Compaq Deskpro 386 is the first of a new generation of industry-standard desktop workstations.
“It’s by far the most advanced high-performance personal computer in the world today. Together, Compaq, Intel, and Microsoft are setting the pace for the rest of the industry with the next generation in general-purpose, high-performance workstations.
“Since the 80386 delivers compatibility with the past and the architecture needed for the future, and since it comes from Intel, there’s no doubt that the 80386 will be the foundation for the next generation of the industry standard.” This last statement seems obvious to many in the audience, but I want to head off any question that something else could possibly be the next-generation industry-standard processor.
Next I describe the specific features of the Deskpro 386, including performance that is two to three times that of 286 machines, increased RAM memory size and speed, increased hard-drive size and speed, and greater flexibility provided by more storage device slots and board expansion slots. I tell the audience, “We have addressed every aspect of the computer and optimized for performance.”
As I finish this part of the presentation, I’m at the point where I would normally wrap up and open the floor to questions. But for this announcement, we’re barely at the halfway point. Still lined up are the leaders of our key industry partners to speak about the significance of the Deskpro 386, the advantages it will provide, and their plans to support it with advances in their own products.
First up is Bill Gates, Microsoft’s CEO. He talks about how excited he is about the Deskpro 386, how Intel, Microsoft, and Compaq have worked together to make it happen, and how Microsoft will support its advanced features with its operating systems. He says Microsoft has Xenix, its version of Unix, available now and that the company is working to bring an advanced version of DOS to market soon.
Next comes Intel’s Gordon Moore. He tells how Intel has worked with Microsoft and Compaq to bring the 80386 processor to market and to achieve full compatibility with its previous generations of processors. He says Intel is capable of delivering a million 386 chips in 1987, meaning there’ll be plenty of them available to meet Compaq’s, and the industry’s, demand.
The Compaq Deskpro 386 is the first of a new generation of industry-standard desktop workstations.
The two final speakers are Ed Asber of software pioneer Ashton-Tate and Jim Manzi of Lotus Development. Both talk about how their companies will take advantage of the capabilities of the 386 with their own applications, and how they’re fully supporting the Deskpro 386.
I return to the podium and tell the audience that Compaq has been working with several large PC customers to test the Deskpro 386 in real user applications. I read quotes from three of them that are effusive in their praise of the Deskpro 386 and the value it will bring to them. One quote, from the Director of Worldwide Distribution at Sisters of Charity Hospital Corporation, says, “The thing we liked best about this machine is that we dropped all the boards we use in it, we dropped all of our software in it, and they worked right from the start . . . [New powerful machines] are only good when they’re compatible.”
If IBM doesn’t pay attention to anything else at our announcement, it should at least listen to this customer. He has succinctly stated the power and importance of the industry standard.
I begin to summarize. “Today, we’re ushering in nothing less than the third generation of the personal computer revolution. Nine years ago, the original Apple II helped start the personal computer revolution. Five years ago, the introduction of the IBM PC was the catalyst for the first true computer industry standard. Today, the Deskpro 386 is bringing us the third generation. For the first time, true 32-bit computing power, within the industry standard, is available in a personal computer. Intel’s microprocessor family is delivering the next generation in performance and architecture. Microsoft’s operating systems are expanding the capabilities of the industry standard. And the applications delivered by Lotus, Ashton-Tate, and others will bring the benefits of the next generation directly to end users.”
Just as the audience might think I’m about to finish, I go into something they’ve never heard before nor expected to hear now: a ten-minute analysis of the potential responses by our major competitor, IBM. “Now I’d like to address a major question raised by today’s announcement. Why has Compaq decided to move out ahead of IBM to the next generation of the industry standard? The answer is that the 80386 is of critical value right now to both the industry and the user community . . . It’s unreasonable to expect the industry and users to wait for IBM, especially if the wait is likely to be a year or more. We have evaluated all the risks and the arguments against moving ahead and have concluded that none of them are valid.
“Let me explain why. First, it’s important to understand that Compaq by itself isn’t trying to set a new standard. Standards are set by users, and the industry standard we have today is in the hands of the user community. The foundation for the next generation of the industry standard is the next member in Intel’s compatible microprocessor family, the 80386. The Compaq Deskpro 386 is simply the vehicle for delivering this important breakthrough to users in the best and safest manner possible through full compatibility with the industry standard. Now, even though this breakthrough is important and useful, and even though Compaq is delivering it safely, what might IBM have to say about all this?”
In the first part of my analysis, I address every component of the Deskpro 386 system and show how we have optimized each area so there is little or no room for IBM to improve upon them. “Even if IBM does introduce some unexpected new capability, expansion bus slots will allow for that capability to be added to the Deskpro 386. So there are really no significant user needs that aren’t being met by the Deskpro 386.”
And finally, I ask the fundamental question—the one I know everyone wants to hear: “What if IBM goes beyond industry-standard 8- and 16-bit slots and offers a new 32-bit slot?” My answer has been thoughtfully prepared and reflects the knowledge gained in developing the first 80386-based PC.
“The only thing a 32-bit slot offers is more speed. We’ve looked very carefully at all the peripheral areas and concluded that the only one that needs the speed of a 32-bit bus is the RAM memory. The Deskpro 386 does provide a 32-bit bus for its memory, and therefore has the maximum speed.
“When you add all this up, I believe it’s clear that a 32-bit port is not needed on this class of product, and therefore it would not have a significant impact on the 80386 desktop workstation market.
“But whatever IBM does, a key factor is when they do it. If six months or more goes by, the 80386 will gain significant momentum without IBM. A lot of users will become satisfied with the advantages of the industry-standard 80386 product. By then, I believe IBM would have a difficult time convincing the majority of the industry to follow it down a proprietary path. The burden would be on IBM to prove its proprietary offering is useful and significant enough to cause everyone to give up the very real benefits of the open environment of the industry standard.”
I pause briefly to let my answer sink in. I feel I’ve distilled the message down to a few straightforward sentences that are easy to understand. Even though the tone of my answer sounds like I have some knowledge of what IBM is going to do, in fact I don’t. I’m simply trying to head off any possibility that a reporter or analyst might form a negative opinion about the Deskpro 386’s chance for success. Subsequent events will indicate that IBM did not pay attention to what I said, although hindsight will eventually show it should have.
I close with a prediction of my own.
“So far, the benefits of the personal computer revolution have been dramatic. Personal computers lead to better, faster decision making. They allow office workers and professionals to do their jobs more quickly and accurately than was ever thought possible. And they give the companies that invest in them a significant competitive advantage. The 80386 generation will bring even more potent benefits than all that’s come before. To those who were involved at the start of this revolution a decade ago, at times it must seem as if things are starting to slow down. Then along comes an advance like the Deskpro 386 and you realize that truly, we’ve only just begun.”
The Deskpro 386 event was a high point for Compaq. Every part of the company had contributed to its success, and many dedicated people had worked long hours in the summer months leading up to it. Afterward, there was little time to rest and reflect, however. The job of convincing the market that the Deskpro 386 wouldn’t be instantly obsolete when IBM introduced its 386 product had just started. We knew there would be many skeptics, so we planned an intensive publicity and advertising campaign for the weeks immediately following the announcement.
Two days after the launch, we ran an eight-page pull-out advertisement in the first section of the Wall Street Journal. The cost was astronomical, but it was part of the price of moving out in front of IBM. Every page contained a bold headline and addressed a key point. The advertisement reflected the same thoroughness we had exhibited in the announcement.
Initial media coverage in the major business publications was generally complimentary of the Deskpro 386, but many contained quotes from analysts, who pointed out the uncertainty of what IBM’s response might be and how it might affect Compaq.
My focus during the announcement on why that wouldn’t be a problem for us had gone a long way toward softening the analysts’ concerns, but they were far from eliminated. The one key point that did sink in was that likely, it would be six to twelve months before IBM would respond. They mostly agreed that our product would see strong demand during the interim.
The media coverage in the trade publications was more effusive with praise of the Deskpro 386. Every article recommended the product for applications that needed more power than the current 286-based products could deliver. Some of the articles went into a detailed analysis of applications where the 386 would make sense.
Media coverage aside, our customers and dealers went crazy over the Deskpro 386. We were already in production at the time of the announcement, but for a while afterward were limited by the number of 80386 chips available from Intel. The supply that was available was due to the “risk wafers” we had paid Intel to start. Our sales team worked with our dealers to make sure that as many 386 machines as possible went to companies evaluating the product. We wanted to be certain that the demand would be there when Intel’s chip supply ramped up in the first quarter of 1987.
The immediate success of the Deskpro 386 was partly due to its ability to run all the existing PC software, plug-in boards, and peripherals. We had established a reputation with the Deskpro 286 for being able to do that, and thus had begun the process of educating the market on the value of backward compatibility. The Deskpro 386 raised awareness of the value of backward compatibility by an order of magnitude because it was such a vivid demonstration of its power. We proved that a company other than IBM could successfully introduce a new, advanced microprocessor as long as the product maintained complete backward compatibility. As a result, many people understood for the first time that it really was an industry standard, and not just an IBM standard. IBM no longer had to be present for the party to continue.
With the aid of a few thousand Deskpro 386s delivered in the fourth quarter of 1986, we finished the year with $625 million in sales, a 25 percent increase over 1985. In most industries, 25 percent annual revenue growth would be spectacular. But for the personal computer industry, it was just mediocre. Overall, PC demand had been weakening most of the year and it was affecting all the PC companies, including IBM.
Rosen had predicted during our 386 announcement event that the Deskpro 386 would usher in renewed growth and vibrancy to the industry. Whether it was the 386, or whether his timing just happened to be right, PCs did see accelerated growth in 1987. For us, there was no doubt the Deskpro 386 accelerated our growth.
The other thing that helped get 1987 off to a fast start was our introduction of the Portable III on February 17. Most of our portable competition was coming from laptops at this point, and we were being criticized for not entering that segment. Our strategy had been to continually decrease the size of our portables, but to make sure they remained “full-function.” The Portable III filled that role well and helped us continue in our portable leadership during 1987.
The Compaq Portable III met our goal of an even smaller yet fully-functioning portable.
Several of Compaq’s executives and I mounted an intense campaign of speeches and interviews during the latter part of 1986 and the first part of 1987 to overcome any resistance to us getting out ahead of IBM. In addition, our major advertising associated with the launch was followed up by an overall higher level of advertising throughout 1987.
Meanwhile, another major event affected us during 1987. In early April, IBM announced its PS/2, the long anticipated and feared “Death Star” aimed at eliminating the clones, among them Compaq.