The “Death Star” Arrives

During the next few days, there were many meetings throughout Compaq to understand all the implications of the PS/2 announcement and to figure out appropriate strategies. Our engineers bought one of the first PS/2 computers and immediately began to tear into it. We also began to study the computer’s hardware and software manuals. Stimac contacted representatives at Microsoft to see what they knew. Barnes contacted employees at Intel to find out if they knew anything they could tell us. Others contacted software developers to determine if they already had applications ready to run on the PS/2, and also companies that sold plug-in boards to find out where they stood.

The IBM Personal System 2 (PS/2) has a new expansion bus called “Micro Channel.”

It quickly became apparent that IBM had done a very good job of keeping details of the new product a secret from the rest of the industry. As far as our inquiries could determine, neither Microsoft nor Intel knew everything about the PS/2 before the announcement.

There were no third-party application programs or plug-in boards initially available or about to be available. If a company wanted to develop a plug-in board for the PS/2, it would first have to buy a license from IBM to use the Micro Channel. IBM was clearly very serious about maintaining proprietary control over its PS/2 and making money from everything that went into it.

In January, IBM had begun sending strong signals of imminent products that would be much more proprietary than its existing PCs. Early in the year, IBM briefed a number of its major customers about products that couldn’t be cloned. Some were told IBM’s next operating system would not run on computers from other companies. IBM had implied its new products would be compatible with its existing products, but would offer advances that could not be legally cloned.

Based on this information, we believed IBM would add a proprietary 32-bit bus to the existing industry-standard 8- and 16-bit buses. This approach made sense to us. But the Micro Channel bus had absolutely no compatibility with the industry-standard bus. That move did not make sense.

We had several intense meetings over the next few days. We realized there was no way to convince the media and analyst communities that IBM had made a mistake.

We decided we needed to convince a significant number of customers to think before they blindly followed IBM down the PS/2 path. To accomplish this, we decided to take a strong position on the negatives associated with the PS/2’s lack of backward compatibility with the existing base of software and plug-in boards. Even if the press didn’t agree with our position, we wanted it to be in print where customers would see it.

We realized we had to be careful not to paint ourselves into a corner and damage our reputation if the PS/2 did eventually become the new industry standard. And we had to be very sure of our facts when we publicly criticized any aspect of the PS/2.

The approach we settled on was to carefully and specifically point out the disadvantages a customer would face when moving to the PS/2. At the same time, we would make it clear that Compaq would always listen to the demands of the market and provide customers with products and features they desired, including PS/2 compatibility. The key was to handle our response objectively and accurately, so that we would not be viewed as defensive or scared.

We were also looking for a way to draw a parallel that would weaken the widespread perception that whatever IBM did was right. The analogy we chose was New Coke, a well-known failure of a market leader to move customers in a different direction. The idea was not to say the PS/2 would fail but to point out it was not a given the new computer would succeed to the same degree that the original IBM PC had.

By mid-April, along with Swavely and others, I began giving speeches and conducting interviews articulating Compaq’s position. We quickly succeeded in getting our message widely communicated. The media loved a conflict, and we became the point man for the opposition in this one. There were so many articles being written that I asked our market research group to send me a report of all the different viewpoints. The following comments occurred in May and were included in the report that I received in early June.

For the most part, our message was understood and well received, but a significant number of people missed part of the message. Adding the analogy to New Coke opened up a hornet’s nest.

Wall Street’s analysts seemed to understand the balanced message best. They followed our logical argument, which established the PS/2’s incompatibility with the existing standard as a negative. But they also understood we were saying we would follow the PS/2 if enough customers demanded it.

Michael Davis of investment advisor Lovett Mitchell Webb & Garrison wrote, “Most important was Compaq’s stated continued commitment to being responsive to customer needs. Over the near term, Compaq will accomplish this by continuing to provide existing and new products that are compatible with the industry standard. Compaq will only introduce products that are compatible with IBM PS/2 when the marketplace demands such products. As long as Compaq maintains this attitude, we see no near-term limits to its growth.”

Michele Preston of Salomon Brothers wrote, “In its most aggressive stance ever, Compaq is promoting the importance of compatibility with industry standards and providing products that provide real—not perceived—user benefits. This will strengthen Compaq’s franchise in business accounts and the dealer channel . . . The company has been meeting with customers, the press, and analysts to make this position clear.”

One who missed part of the message was Peter Labe of Drexel Burnham Lambert when he wrote, “Canion’s recent speech is both brilliant and articulate. It is as persuasive a case as we have seen for a two-standard market—the IBM standard (as personified by PS/2) and a de facto standard of all the existing stuff of which Compaq is the most ardent exponent. We believe there is great risk this time that Mr. Canion might not walk on water. He is right technically, no doubt—but in actual practice, we will bet on IBM. In the history of the compatible industry, where there is a history since 1969 in other products, it has been conclusively proven that there is no such thing as ‘almost compatible’. . . . The fact of the matter is that IBM has regained the initiative and leadership position.”

Not surprisingly, executives from some large companies completely toed the IBM line. In response to a query from a Compaq sales representative, a senior executive of ARCO told us, “You folks are making some very big assumptions. . . . IBM sets the standards, IBM drives the standards . . . and if IBM wants to it can lock out all PC vendors.”

There was also relatively strong support in the PC industry press for the infallibility of IBM. A PC Week editorial on May 5, 1987, stated, “Users are hearing a lot of rhetoric in the aftermath of IBM’s PS/2 announcement. . . . Rod Canion publicly compared the PS/2 to Coca-Cola’s well-publicized marketing disaster, ‘New Coke’ . . . the implication is that PS/2 is a radical and unnecessary departure from users’ current installations. The view . . . is simplistic. Worse, it creates unnecessary anxiety for users who feel pressured to choose between allegedly irreconcilable architectures.” He was partly correct. We were trying to get customers to think things through and not just blindly follow IBM.

Most disappointing to me was the reaction of leading industry consultants. Stewart Alsop of the PC Letter wrote, “Taking Compaq’s statements at face value indicates that it will stick with what it has and take a different route than IBM. That particular strategy is an extremely dangerous one, if you look at the history of the computer industry: the only companies—without exception—that have prospered in the long term are ones that have either stuck with what IBM has defined as the standard (like Amdahl) or ones that have departed completely from the standard (like Digital Equipment and Apple).”

Aaron Goldberg of International Data Corp., a market research firm, wrote, “Rod’s approach is obstinate . . . outlook for Compaq is trouble if they keep thumbing their nose at IBM. . . . Verbal commitment to the new standards is important. . . . Ostriches need not apply. . . .”

Doug Cayne of Gartner Group wrote, “Compaq publicly denying the need for new architecture elements . . . privately working hard to imitate them . . . biggest risks are pricing and arrogance.”

I decided the analysts didn’t understand the risk associated with following IBM to the PS/2. In spite of the confusion and mixed reactions, we stayed with the same message. It was getting through to customers and seemed to be having the desired effect, since our sales continued to grow through the summer.

During all this, we met many times to discuss how we should deal with the PS/2 and Micro Channel from a product development standpoint. Our engineers concluded there were no real advantages associated with the Micro Channel. They reported that each of Compaq’s industry-standard products outperformed the most comparable PS/2 product.

Adding this information to our speeches and interviews strengthened our arguments and led us to decide we had to focus part of our engineering team on finding more innovations that would continue to give our industry-standard products performance advantages over IBM’s PS/2. We believed the tide could turn quickly if new PS/2 products were able to outperform ours. But we could not escape the possibility IBM’s brand and marketing power might be strong enough to actually pull it off—even if we outperformed it.

We were sure many of our competitors were rushing to develop new products compatible with the PS/2, even though they had to sign a license with IBM and agree to pay very significant royalties. If the industry moved to a new standard, and if our PS/2-compatible products were behind our competitors, Compaq would almost certainly lose its leadership position. So we decided we would have to carve out a significant part of our product development resources and get started on reverse engineering the PS/2. This was a very difficult and particularly distasteful decision for me, but we were pragmatic enough to realize that waiting too long to get started could lead to disaster.

Even before our official decision had been made, a small group of engineers had started the process. After dismantling an early PS/2 to understand exactly what it was, they began to plan an approach.

We discovered a lot more resources would be necessary to get the job done this time. The same people who had reverse engineered the original IBM PC were still around, but IBM had taken more time developing the PS/2 and it contained several special customized chips called “ASICs,” or application-specific integrated circuits, which would be very challenging to reverse engineer.

We were stunned by an estimate of the people and money it would take to complete the process and design a product we could take to market. The bottom line: We were going to have to essentially split our engineering resources down the middle, half working on PS/2-compatible products and half developing industry-standard products that would enable us to stay ahead of IBM in performance. Two years earlier, this would not have worked. But we had steadily expanded our engineering and software teams and would be able to do a reasonable job of both, at least for a while.

It didn’t take long for this decision to leak to the industry. From the beginning, we had stressed to employees the importance of keeping product plans secret, but there was just too much attention being paid to what we might or might not do for something this significant not to get out.

We were stunned by an estimate of the people and money it would take to complete the process and design a product we could take to market.

Included in the report I received in early June was an excerpt from an article that appeared in the June 2, 1987, edition of PC Week: “Compaq hedges its bets . . . straddles the fence on issue of PS/2. Bolstered by Compaq’s burgeoning sales, Mr. Canion has been uncharacteristically strident in his criticism of the PS/2 and has eagerly assumed the mantle of standard-bearer for PC compatibility. Although he speaks from a position of strength, Mr. Canion’s anti-PS/2 remarks startled many industry observers. The contrast between Mr. Canion’s avowal and his design team’s feverish quest [to duplicate the PS/2] illustrates the ambivalent relationship of Compaq to IBM and the parallel strategies the company must pursue to maintain its leadership position in the industry.”

I was disappointed that there had been a leak, but glad the article had gotten it right. I thought it might actually have a positive effect on those who had missed my comments about following the PS/2 if enough customers demanded it.

I felt strongly we needed to continue to introduce products that clearly outperformed the comparable PS/2s from IBM, thereby supporting our position that there was no real performance advantage with the Micro Channel. And I really wanted to make the point there was more performance to be gained from our technology advancements than from the Micro Channel. My chance came with the product we announced in September 1987.

In the process of working with Intel to get the first 386 to market in 1986, we had mainly focused on making sure the chip was totally compatible with the 286. During that process, our engineers learned the performance of the 386 was severely limited because it couldn’t access the memory and Input/Output buses in parallel. In other words, if an operation with a peripheral was in progress, the processor couldn’t access memory until the I/O bus was free.

Paul Culley, our leading hardware architect, had begun working on a way to fix this bottleneck as soon as the Deskpro 386 had gone into production. He came up with a creative solution that was being implemented in the next product, the Deskpro 386/20, with the “/20” signifying a 20-megahertz version of the 386 chip.

A memory cache controller took over both the memory and peripheral buses and accessed them in parallel. In addition, while Intel had a good math coprocessor for the 80286 called the 80287, it didn’t have one ready for the 386 when it was going to market. We had used the 80287 as the math coprocessor for the Deskpro 386, but it was far from the optimal solution. Since Intel hadn’t caught up with a viable 80387 chip, we chose to work with the chip-design company Weitek to use its math coprocessor chip in the Deskpro 386/20.

The result of these two innovations was that the Deskpro 386/20 would run more than 50 percent faster than other 20-megahertz 386 machines. And since the PS/2 Model 80 used a 16-megahertz 386, the Deskpro 386/20 would run almost 100 percent faster. That was the kind of clear performance advantage I wanted to be able to demonstrate.

On September 29, 1987, we announced two major new products, the Deskpro 386/20 and the Portable 386, which also used the 20-megahertz 386. In order to highlight the unique technology we had invented, we created the name “Flex Architecture” to refer to the combination of features that resulted in such amazing performance advantages. The announcement touted the performance of both products, and the demonstrations held afterward clearly showed them as cutting-edge.

In what could have easily been a ho-hum announcement of a clock speed upgrade, we succeeded in making a very strong statement about our leadership in PC technology. Our timing was nearly perfect, since the first PS/2 386 machine had just started shipping. We were able to back up our strong performance claims with simple demonstrations comparing the Deskpro 386/20 and Portable 386 with the IBM PS/2 Model 80. The performance differences were so clear they couldn’t be ignored or explained away. We really were far ahead of IBM in delivering high performance, and we did it with a machine that ran all the existing software and all the existing plug-in boards as well.

Most of the press and some of the analysts got the message we were delivering. In fact, the coverage of our two new machines was the most positive we had ever received. In addition to performance, two other aspects really seemed to make an impression. One was our name for the “secret sauce” that made our machines so much faster. By calling it “Flex Architecture” and referring to it over and over by that name, it began to take on an air of significance similar to IBM’s Micro Channel. Although IBM had continued to tout the performance advantages of Micro Channel that were “coming soon,” we demonstrated that our performance advantages were “here and now.”

The other thing that made a strong impression was that we were able to put a machine with all this performance into a portable package that weighed just over 20 pounds and was much smaller than the original Compaq Portable. As we stated in our advertisements, it was “pound for pound the most powerful computer on the planet.” This was one more example of Compaq doing something no other company had ever done. The positive impact on our brand image and reputation was simply beyond measure. Compaq was truly becoming the darling of the press and of Wall Street.

BILL RETIRED FROM COMPAQ in 1987. He was the first founder to leave the company, and I was sad to see him go. Jim and I had been right when we decided in 1982 that Bill’s contribution would be as important as our own. Bill’s recommendation to sell only through IBM authorized dealers and his leadership of the sales and marketing teams was absolutely critical to our success. He was also a mentor to Swavely early on and later to Ross Cooley, an ex-IBMer who developed into a great vice president of sales after Bill left.

As 1987 drew to a close, we had little time to dwell on how amazing a year it had been. In achieving over $1.2 billion in sales, we set another record for the fastest any company had ever reached sales of a billion dollars. And we had done it in the same year our most dangerous competitor, IBM, had made its highly aggressive move in the PC market. Most of the world still did not understand how we did it, but they couldn’t deny we were thriving in spite of IBM’s repeated attacks that were expected to stop us.

The Compaq Portable 386—“pound for pound the most powerful computer on the planet.”

In spite of our excellent financial performance, though, I continued to feel like there was a dark cloud hanging over us, and its name was Micro Channel. We met often to discuss the strategic issues it created, always using “The Process” to try to come up with a solution. But this problem was the toughest we had ever faced, and we really weren’t making much progress. I was certain that if we were going to solve it, we were going to have to ask ourselves the right question.