Intel Begins to Drift
MARCH 15, 1989, 10:30 A.M.
Hugh Barnes, vice president of engineering, has requested an emergency meeting with our strategy team. As soon as everyone is seated, he starts. “I believe we have a serious problem with Intel. The 486 chip is behind schedule and falling further behind.”
I ask, “What’s the problem? Is the chip becoming too complex?”
“That’s part of the problem. But the real issue is, they don’t have their best people working on it. They’ve started a project to design a RISC (Reduced Instruction Set Computer) processor and assigned several of their most experienced chip designers to handle that instead of the 486. Apparently they’ve prioritized the RISC project ahead of the 486.”
I shake my head slowly. “What are they thinking? How could getting into the tiny market for RISC chips be more important than protecting their monopoly with the industry standard?”
“I don’t know. Maybe Scott McNealy’s rhetoric is affecting them.”
The CEO of Sun Microsystems is getting a lot of press with his pronouncements about how much faster his SPARC RISC processor is than Intel’s CISC-based, X-86 processor (each Complex Instruction Set Computer instruction can execute several low-level operations). He’s been saying that the next-generation SPARC will be two to three times faster than Intel’s next-generation 486.
“Is there any truth to McNealy’s claims?”
Barnes thinks for a moment. “There’s some speed advantage to the RISC approach, but I doubt it’s as much as he claims.”
I look around the room. “It appears we’re going to have to convince Andy Grove [president of Intel] that the 486 is more important than chasing the RISC market. Does anyone have an idea about how we can do that?”
Stimac, vice president of systems engineering, speaks up. “I think we can get Bill Gates to help. He has a vested interest in the industry standard staying on top.”
I reply, “OK, contact Gates and see if he’s willing to help. Hugh, see if we can get a meeting set up with Andy. I can fly out there, but I think we should meet somewhere off-site from his office. And let’s do this as soon as we can.”
I sit there and think for a minute. I’m wondering why Intel doesn’t understand the power of the industry standard and the importance of keeping it at the forefront of performance. As if fighting off IBM wasn’t enough, now we have to try to keep Intel from letting the 486 chip slip further behind. We can’t afford to lose the incredible advantage the industry standard is giving us.
I realized the meeting with Grove would be pivotal for the future of Compaq and the open industry standard. I had to find a way to get him to fully understand and appreciate the open industry standard’s strengths and benefits. Then it hit me: While I had talked about the importance of the industry standard for many years, I had never fully articulated all its benefits and how they worked together to deliver such a significant impact. I had to get the benefits straight in my own mind before trying to convince Grove.
I spent some time writing down my thoughts, and then arranged them in order of importance. I looked at my completed list and thought it was compelling:
1. guaranteed access to the latest software and peripherals
2. rapid introduction of key innovations and advanced technologies—higher performance, higher capacity
3. lower prices—continual downward pressure due to intense competition
4. broadest range of brands and product choices ever
5. ability to use existing software and plug-in boards when upgrading, even to a different brand PC
The combination of all these working together was stronger than any one individual benefit, and gradually created a chain reaction in the PC market. As prices went down and performance went up, more and more users were drawn in. Once they were hooked on the increased productivity they achieved, they were often attracted to upgrade to new computers with more performance and innovative new features, especially since upgrading was such an easy process.
As the rate of market growth increased, Intel and Microsoft began to generate incredible profits, which they wisely plowed back into rapidly developing advanced technologies. As these came to market, more users were drawn in, and then the cycle repeated.
These benefits created such a strong attraction to industry-standard PCs that even IBM couldn’t easily pull customers away. However, if Intel couldn’t deliver a high-performance 486 processor on time, it would give IBM, Sun Microsystems, and others the opportunity to tear down the open-industry-standard empire.
APRIL 4, 1989, 11:00 A.M.
I knock on the door of a small hotel suite in Silicon Valley. Bill Gates opens the door and invites me in. Already in the room are Andy Grove and Gordon Moore, president and chairman of Intel respectively. The four of us exchange greetings and then sit down around a table.
I jokingly ask, “Who called this meeting?”
Grove looks at me seriously and says, “I think you did.”
“Well, I guess I did. Thanks for coming. I’ll get straight to the point.” I look directly into Grove’s eyes and say, “My team and I believe you are giving a higher priority to developing a RISC chip than to completing the 486, and the result is that the 486 is slipping further and further behind.”
Grove is ready with a response. “Both chips are very important to Intel. We have excellent teams working on both. The 486 has had a few problems, but it’s back on track.”
I reply, “Look, Andy. I’m not going to try to tell you how to run your business. I know getting into the RISC chip business is important to you. But as you know, Compaq’s engineers have worked closely with yours for a long time now, and they’re telling me your very best people are working on the RISC chip. What I want to do is make sure you understand how important the 486 is to all our futures.”
Grove comes right back at me. “The microprocessor market is much bigger than just the X-86 segment. We have to be able to develop more than one new product at a time. The problem is McNealy’s hype seems to be working, because we see a lot of computer companies evaluating the Sparc chip. If we don’t provide those companies with a viable alternative soon, Sun will end up dominating that business. And with the performance advantages RISC chips offer, eventually they’ll win out over X-86-based PCs.”
I think about this and then ask, “How big is the performance advantage?”
“It depends on what new performance enhancement tricks we can come up with for the X-86. The RISC clock rate will always be higher, so we have to find other ways to improve our performance. We think it’ll end up somewhere between 30 percent and 100 percent.”
“Hmmm, that much. Well, obviously the smaller the gap, the better. But I believe that the advantages a customer receives by using an industry-standard PC are significant enough to offset a lot of performance. When we introduced the first 386 PC, it was accepted so quickly because it could use all the existing software and peripherals. Sparc workstations won’t even come close to having that much flexibility. Some users who have extreme performance requirements might switch, but most users will continue to buy industry-standard PCs.”
I pause to let that sink in, and then continue. “Think about it this way. When the PC market first started, there were several microprocessors in the race. Then IBM comes along and gives us all a gift by introducing a PC that can be cloned. And as a result of all our efforts, an industry standard develops and becomes dominant. The X-86 is a required component of the standard, which means Intel has won that race. As long as you continue to deliver competitive performance advances, you can’t be removed from the throne. But if you make your top priority getting to a RISC chip that is competitive with the performance of the Sparc chip, it’s like starting a new race where you may or may not be the winner. You should be doing absolutely everything you can to protect the industry standard and keep the X-86 advancing as rapidly as possible. You should protect your effective monopoly on the industry standard instead of making a new race your top priority.”
I pause again and watch Grove think about this. After a while, he looks at Gates and says, “How much of a performance disadvantage can we have before a significant number of end users begin to switch over to Sparc?”
Gates replies, “A lot of the actual performance depends on the software, and it’ll vary by the type of application. For most applications, a 30 to 50 percent performance disadvantage won’t be a problem. I agree with Rod. As long as they don’t have a sustained advantage of well over 50 percent, I don’t think we’ll lose many customers. And the ones we lose will be focused on a narrow range of applications in the workstation arena.”
Grove adds, “The performance differential isn’t a constant thing. When we come out with a new chip, most of the time we’ll pass up their existing chip. Then when they come out with their next chip, it’ll probably pass ours. So it’s not like we’ll be behind them all the time.”
The four of us continue deeper into the details of performance and advantages of the industry standard. I try my hardest to explain why the industry standard has become so powerful. From my perspective, the pendulum swings in favor of my position for a while and then swings back in the other direction. As the meeting winds down, it is still unclear to me whether or not Grove will change his priorities.
Finally, Grove says, “I just can’t get comfortable accepting the position of being at a performance disadvantage to a major competitor. Logically, I agree with your arguments. But I need to think about this and discuss it with my engineers. We can’t afford to get this wrong.”
I reply, “You’re exactly right. None of us can afford to get this wrong. Please call me at any time of the day or night if you want to discuss it more.”
As we leave the room, we all shake hands and agree to follow up soon.
When I arrived back in Houston, I still did not feel sure about what Intel was going to do. It was very uncomfortable to not be sure of something that could affect our company so much. I met with our strategy team shortly after returning to the office and told them how the trip had gone. The mood was somber.
Five weeks later, Barnes bursts into my office with a big grin on his face. He had just heard that Intel was reorganizing its processor development teams. Not only were they putting their top engineers on the 486 project, but they were also starting development on the 586 chip, which would eventually become known as the Pentium. Intel was continuing to work on a RISC processor, but not at the expense of either of the industry-standard chips. This was the best news that I could have imagined. I was thrilled as I contemplated what this would mean for Compaq’s long-term success.
WE CONTINUED OUR TORRID PACE of new product announcements in 1989 with the introduction of the Deskpro 386/33 in May, two months ahead of IBM’s comparable machine. As usual, we began shipping the product to our dealers ahead of the announcement. While several of our competitors, including IBM, had begun announcing new products several months before their shipments actually began in order to be viewed as “first” and as the “technology leader,” none of them succeeded in getting large numbers of their new products to market ahead of us. Plus, the press had become wise to their practice.
The 33-megahertz 80386 chip was the last version of the 386 introduced before the 486 chip took over the performance leadership position late in 1989. We had succeeded in maintaining our performance leadership in 386-based computers throughout the entire 386 era. An article in Computer Reseller News on May 29, 1989, written by Alice Greene titled “Compaq Unveils ‘A Screamer’” declared, “33 MHz System Cements (Compaq’s) Lead In 386 Market.”
On July 14, we signed a patent cross-licensing agreement with IBM. The company had first contacted Bill Fargo, our general counsel, to begin discussions shortly after the PS/2 announcement, but not much progress was made until September 1988. Then, just after the EISA announcement, IBM seemed to get upset and notified us that we were infringing many of its patents. At first I thought it was planning to use the tactic TI had used on us in 1982 and try to slow us down with a lawsuit. Instead, it was more like part of a larger program to extract some value for its intellectual property that competitors were using. It turned out that IBM was also infringing some of our patents. It was particularly concerned about our Roberts patent, which covered the dual-switching monitor Ken Roberts invented in 1982. David Cabello, our in-house intellectual property attorney, negotiated with the company for almost nine months before we finally reached an agreement. We paid IBM a one-time fee of $130 million that gave us unlimited access to all their patents filed through 1993. And it was much smaller than it would have been if we hadn’t put a lot of effort into filing for patents on our own inventions.
Then, on October 15, 1989, we raised the bar for laptop computers by introducing the first “notebook” computer, so named because of its 8½ x 11-inch footprint, the same dimensions as a piece of notebook paper. In spite of their small size, the Compaq LTE and LTE/286 were full-function personal computers that employed several unique advances in technology. One of these, the first use of a 2½-inch-wide hard drive, enabled us to deliver up to 40 megabytes of storage.
We had made a strategic investment in Conner Peripherals, the company that invented this size drive and, as a result, were able to secure a six-month exclusive period on the tiny new component. We also received a six-month exclusive period for each new, higher-capacity drive introduced during the next three years. Thus, having defined a new category of laptop computer, we effectively owned that market segment for several years. As a footnote to history, Apple employed a similar strategy for its early iPods in 2001 and gained comparable results.
On October 17, 1989, an article in the New York Times written by Peter Lewis titled “Compaq Does It Again” opened with “Every once in a while a new computer comes along that leaps ahead of other machines in its class. This time, once again, it comes from the Compaq Computer Corporation.”
True to our fundamental portable product philosophy, we had waited until the technology enabled us to deliver a “full-function” notebook computer. Contrary to analysts’ predictions that we had missed the laptop market, the LTE and LTE/286 added to the strong start provided by our earlier SLT/286 and solidified our leadership position in the overall portable market.
One of the most visible signs of the rapid advance of technology in the ’80s was the shrinking size and increasing power of portable computers. Every part of the computer changed dramatically over that seven-year period, and due in large part to the industry standard, many different companies contributed to those advances. The opportunity created by the rapid growth of the industry-standard segment of the PC market attracted most technology companies and provided the incentive for them to push themselves to beat their competition. And since they didn’t have to adapt a new component to five or ten different systems, they could focus on the next advance while reaping the rewards of the previous one in the giant and rapidly growing industry-standard market.
Evolution of Compaq Portables 1983–1989. The original Portable, Portable II, Portable III/386, SLT/286, and LTE/286.
The most important announcements of 1989 were yet to come. At the time of the original EISA announcement in September 1988, we had committed to launching our first EISA products by late 1989.
We would make good on that commitment, but there was a slowdown in the PC industry during the second half of 1989 that forced us to announce lower-than-expected earnings in the fourth quarter. The statement was released just a week before the launch of our new EISA products. We had become the darling of Wall Street with our financial performance over the last three years, and Wall Street did not kindly treat companies that let it down.
We had become the darling of Wall Street with our financial performance over the last three years, and Wall Street did not kindly treat companies that let it down.
During the entire week before the announcement, the press was full of reports about our stock price getting battered. That really put a damper on the environment surrounding our upcoming product launch. This wasn’t at all what I had in mind, but I pressed on, believing the announcement would be strong enough to put the earnings disappointment behind us.