"This is a very bold move for us," said Rich Marcello, vice president and general manager for Compaq's performance systems group. He noted that the Alpha road map extended for several more years, but that starting around 2004, when the chip would have competed head-to-head against Intel's Itanium, its performance advantage would begin to erode significantly. "When we looked at our road map and overlaid it with Intel's road map, we found there was no substantial performance benefit," he said. "Basically, we are saying that we couldn't differentiate ourselves at the CPU level."

Compaq currently has some 500 microprocessor engineers and spends about $150 million annually to develop its internal processors. Effective immediately, 200 of those designers will go to work for Intel, where it is expected that they will be able to implement their Alpha expertise into future iterations of Intel's Itanium family. Eventually, the rest of Compaq's MPU designers will become Intel employees. Marcello said the company will be receiving some significant financial compensation from Intel, but he would give no details.

Until now, Compaq, Sun Microsystems Inc. and IBM Corp. were the only system vendors that designed their own processors in a bid to compete against Intel. Marcello said that his company's decision was a reflection of the changing face of the processor world, where the CPU has become a commodity, and he predicted that the others would follow Compaq's lead. "More than 95 percent of the customers I've talked to said this is the right decision for us to make," he said. "Eventually, I believe that Sun and IBM will have to do this as well."

However, executives at Sun dispel this notion. Sue Kunz, director of marketing at Sun (Palo Alto, Calif.), made it clear that Sun has no intention of exiting the processor arena. There are similarities between Sun and Compaq, as both design powerful processors that are used at the heart of their own high-end servers and workstations, but Kunz points out an important difference. "I believe that we have always done a much better job than Compaq," she said. "We have always built our own chips, software and systems. Building all the parts together is a key part of what makes us Sun."

She praised the Alpha architecture but added that its absence from the market could be a major opportunity for her company. "You can't just swap out a processor and have the systems run exactly the same. This will be felt by the customers," she said. "If I was an Alpha user, right now I'd be sitting down and trying to figure out my strategy going forward. And at Sun, we're going to be calling all of them. We've got feet out on the street right now."

Compaq expects to shift its entire line of 64-bit servers and workstations to the Intel architecture Itanium family by 2004. This will include both the Alpha-based systems and the MIPS-based Himalaya family of products. Until those first Itanium-based systems are available, it will support both technologies, and new versions of both are expected in the near future.

Compaq is currently working on the EV7 version of the Alpha chip, which is expected to debut in the company's Marvel line of systems in 2003. A subsequent version of that design will also be released, which will simply incorporate a process shrink. Marcello said that these parts, and existing Alpha designs, will continue to be produced as long as there is demand from existing customers, which he said could continue through 2012. "These will be the last Alpha designs," he added.

Compaq acquired the processor architecture in 1998, with the $9.6 billion acquisition of Digital Equipment Corp. As part of the complicated terms of this deal, Samsung Semiconductor obtained manufacturing rights for the chips and still serves as a foundry for Compaq's low-end Alpha products. IBM produces the more advanced Alpha parts, and both companies will continue to produce the chips, said Marcello. Compaq acquired Tandem Computers in 1997 and its MIPS-based processor team.

Steve Leibson, editorial director of Microprocessor Report, said he was not surprised by Compaq's decision. "It was just a matter of time before Compaq abandoned Alpha," he said. "I call this deal 'Itanium eats Alpha.'"

He noted that the company has been developing parallel 64-bit strategies, around both Alpha and MIPS, which, from a marketing point of view, was hard to reconcile. While both of them offer solid performance, that is not always enough to support a product line. "When I look into the future, I see 20 or 30 other server and workstation vendors already jumping on the Itanium bandwagon. Sun, IBM and Compaq are the last holdouts, and the question is whether they want to be one of the three last holdouts. It is very difficult to go it alone, and there is a limited amount that one company can do against the rest of the world."

However, Leibson does not see either Sun or IBM following suit, as both have managed to develop very healthy businesses around their chips. Sun develops the UltraSparc line, while IBM markets the PowerPC design.

"Compaq decided, correctly, that they are a system company and not a processor company," he said. "There are a lot of good architectures that have fallen by the wayside, and the industry can't support more than a few main architectures. This is pretty much the death knell for the Alpha."

Marcello said that while the decision is the right one, it was also a difficult one. "We were proud of our ability to design processors," he noted. "But we have to be able to let it all go."