Exceeding Expectations
NOVEMBER 6, 1989, 11:00 A.M.
More than 1,000 people are gathered in the Houston Astrohall for what they think is Compaq’s launch of our first 486 PC incorporating the EISA bus. The program starts out like one of our typical major product announcements, with Chairman Ben Rosen giving opening remarks.
“Today Compaq, the feisty company that has continued to recreate itself for seven years, is recreating itself once again.”
As he completes his introduction, the hall is darkened and a dynamic laser light show complete with smoke and blaring music begins. The laser show subsides, then up comes a front screen filled with bold lettering: “UNLEASHING THE POWER.” And on center stage is not a new PC like the audience is expecting, but an entirely new type of computer, the SystemPro.
I step to the podium. “The Compaq SystemPro combines a flexible system processor design enabling the user to take advantage of both 386 and 486 processor technology, including the capability to utilize multiple processors for unmatched performance; the 32-bit Extended Industry Standard Architecture expansion bus; Compaq Flex Architecture with Multiprocessing Support; and innovative, new, fixed-disk-drive array storage technology.
“As the industry leader in high-performance 386-based personal computers, Compaq is in the best position to migrate that technology to meet the expanding demands of departmental computing, networking, and multi-user applications. The SystemPro delivers superior performance at better prices than most minicomputers available today. We believe that its exceptional power and expandability make it a strong beginning for an entirely new class of personal computers, the PC system.”
I list the details of each of its features, then describe the new software from Microsoft, SCO, Novell, and Banyan that is necessary to manage this advanced hardware in networking and multi-user environments.
We are aware that for many in the audience, details about “bits” and “bytes” are confusing. So I shift gears and begin comparing the SystemPro’s price and performance with specific minicomputers sold by our major competitors. For the comparison, each computer was configured to serve as a host to sixty terminals in a multi-user environment using the UNIX operating system. I tell the audience that in comparison with the HP 9000 Series 835 minicomputer, the SystemPro is three times faster and costs $68,000 less, and it is six times faster and $135,000 less than DEC’s VAX6310.
After finishing with the SystemPro, I introduce the DeskPro 486/25, Compaq’s first PC based on the new Intel 486 chip. I explain in detail its features that are oriented toward advanced applications such as computer-aided design (CAD) and engineering workstations (EWS). The 486 delivers up to three times the performance of a 386 running at the same clock speed, which clearly moves it into competition with dedicated workstations.
I end by telling the audience that, unlike our normal practice, the 486 systems won’t be available until early in 1990. During our testing of early production models, Compaq engineers discovered a bug in the 486 chip that Intel is now in the process of correcting. I didn’t mention that one of our major competitors, in attempting to beat us to market, had already begun shipping 486 PCs with the bug in them.
After the presentations, attendees move to another room, where Compaq employees are demonstrating various networking and workstation configurations on more than a hundred computers. A significant number of staff from software and peripheral companies are also demonstrating their EISA products. There are even some IBM PS/2 Model 80s set up for comparison to the SystemPro. Demonstrations continue for several hours while attendees absorb the significance of these advancements.
The Compaq SystemPro, our first product using EISA, outperformed minicomputers at much lower prices.
We had gone beyond the obvious. As we had done so often, Compaq’s strategy team had asked the right question: What customer needs can a computer with the EISA bus address? We realized that EISA could take the performance and capabilities of a 386- or 486-based computer into applications that PCs previously had not been able to address.
We began to look at functions traditionally performed by minicomputers and concluded that with new, higher-performance network interface controllers and disk-drive array controllers taking advantage of the increased performance capabilities of the 32-bit EISA bus, a 386-based computer could replace minicomputers in many workgroup applications. With a 486 processor, performance actually went beyond most minicomputers. And just to be absolutely sure we delivered the highest-performance EISA product, we decided to include an optional second processor that could be either a 386 or a 486.
We were aware of the problems IBM had created when it introduced the Micro Channel without any add-in boards available to demonstrate its performance potential. That’s why we were committed to delivering a complete system, from Day One, that clearly showed the amazing performance EISA made possible. Compaq engineers worked with network hardware companies such as Novell to develop intelligent network controllers using the bus-mastering capabilities of EISA.
Other Compaq engineers who specialized in storage technology developed a disk-drive array controller that delivered four times the performance of non-arrayed drives, and therefore took advantage of the speed of the 32-bit EISA bus. After we combined dual processors with the intelligent network controller and the disk-drive array, we had a system that could win benchmark comparisons against minicomputers from HP, DEC, IBM, and others.
When I presented the results at the SystemPro announcement, it was almost too much for the audience to believe. Apparently, no one had ever seen a comparison between microprocessor-based systems and market-leading minicomputers. At first, it must have seemed like smoke and mirrors. We had anticipated skeptical reactions to our numbers and had hired the widely respected performance testing firm of Neal Nelson & Associates to conduct comparisons.
News reports of Compaq’s new product announcements were generally positive, but somewhat muted due to our earnings warning the previous week. We were still projecting about 20 percent year-over-year quarterly revenue growth and net margin consistent with our model of 8 to 10 percent. The issue was simply missing expectations, but it was enough to take some of the gloss off our reputation for consistently beating expectations over the last three years. There was still some confusion over which bus would win in the long term, and now Compaq was adding another confusing claim that these new products could outperform minicomputers while costing much less. Plus, some people just couldn’t accept that IBM had been beaten so badly.
A few reports did get the story right, however. Peter Lewis in the New York Times on November 7, 1989, wrote a story titled “Compaq Redefines High End.” He noted how we “introduced two new products that redefine the high end of personal computer performance and are likely to change the landscape of office computing in the 1990s.” Then, after reporting on the details of our new products, he concluded, “The debate over buses is a popular pastime in the engineering shops of IBM and its rivals, but we have yet to hear anyone say, in real life, ‘Gee, that Micro Channel sure has made my life easier.’ We suspect that most users do not care whether the data bus is ISA, MCA, the new EISA, or COAT (Chipmunks on a Treadmill), as long as the job gets done. The real effect of today’s Compaq announcements will be felt in years to come, as desktop systems connected in networks replace the hulking minicomputers and mainframes that have been the backbone of office systems for 20 years.”
Lewis was probably right; most users didn’t care about the bus, but a lot of big companies did. If they stayed with the industry standard, they needed to be sure they would have access to the most advanced technology as it came to market. Conversely, if they were going to have to switch to Micro Channel–based PCs eventually, they wanted to minimize their investment in computers that would soon become obsolete. With the SystemPro announcement, we made it very clear that if they stayed with the industry standard, they would have access to advanced technology when they needed it.
In retrospect, none of the press or analysts saw the real significance of the SystemPro announcement. We had gone well beyond their expectations before, so it was easy to view the announcement as another event where Compaq had introduced a better product than their competition.
But this time we went far past merely exceeding expectations—this time, Compaq had called IBM’s bluff and won. We proved that IBM had misled its customers and the industry when it said it would have to give up compatibility to get advanced performance. IBM did it again after the EISA announcement when it said it had looked at what the “Gang of Nine” had done and rejected it because it couldn’t deliver the performance of the Micro Channel.
IBM had spun and stretched the truth all along for the obvious purpose of reducing or eliminating competition, but it could no longer get away with it. The SystemPro delivered technology advances way beyond anything that had even been discussed for the PS/2.
The SystemPro announcement effectively eliminated the only remaining logical argument for shifting to the Micro Channel. In the two and a half years since the original PS/2 announcement, essentially all of the customers who were going to shift to the PS/2 because of IBM’s brand had already done so. It was clear to everyone that the Micro Channel part of the market was going to have higher prices because IBM had essentially eliminated any real competition. Customers also realized that once they had gone through the pain and expense of making the shift to Micro Channel, shifting back would be difficult. So if they were going to have to pay more for their PCs going forward, there needed to be a good reason for making the shift.
That reason had been, according to IBM, there would soon be desirable capabilities available on the Micro Channel that would not be available on industry-standard PCs. Since no such capabilities had emerged during the last two and a half years, many people were left wondering. But when Compaq demonstrated real, useful capabilities with EISA on the SystemPro that eclipsed anything available on the Micro Channel, the last significant reason for shifting to Micro Channel was eliminated.