PARK RIDGE, Ill. — Intel Corp.'s impending sale of its Ziatech unit for less than one-fiftieth of the unit's purchase price has sent jaws dropping as industry players and analysts divine whether the move signals a telecom migration away from the CompactPCI platform Ziatech pioneered.

Telecom's struggles notwithstanding, the sale of the Ziatech portion of Intel's Communications Platform Division (San Luis Obispo, Calif.) to Performance Technologies Inc.for $3.8 million — a staggering slide from the August 2000 purchase price of $222 million — sucker-punched the industry. "No one expected the selling price for Ziatech to be as low as it is," said Eric Gulliksen, project director for the Embedded Development Group at Venture Development Corp. (Natick, Mass.). That has industry watchers asking why.

The sale, to be completed Monday (Sept. 30) or Tuesday (Oct. 1), could be a sign that CompactPCI boards are in as dire straits as the telecommunications industry they supply, or it could merely mean that Intel has squeezed what it needed from Ziatech and now is moving on to other things, analysts said.

Electronics executives said much of the embedded market was caught off guard because at the end of 2001, CompactPCI still appeared to be booming. The technology, originally developed for embedded systems as a more robust alternative to desktop PCI-based systems, grew quickly between 1998 and 2001. In January, on the eve of the Bus&Board/2002 Conference, representatives of the PCI Industrial Computer Manufacturers Group (PICMG) declared that CompactPCI had become a $1 billion-a-year industry. The market was blossoming at a 40 percent annual growth rate, they said, and one analyst predicted that it would reach $1.4 billion by 2004.

Analysts now say, however, that the CompactPCI market — which is strongly linked to the fortunes of the telecommunications industry — took a more serious blow than observers first realized during the general economic slide that began in 2001. "As little as a year ago, or even eight months ago, no one foresaw telecom being this bad," Gulliksen noted.

Intel's low selling price for the Ziatech unit represents a stunning reversal of fortune and reputation for the company that invented CompactPCI. Competitors still say that Ziatech was regarded as a leader in the industry and that its products were considered state-of-the-art prior to its acquisition by Intel in 2000.

"Ziatech was always one of the best promoters of CompactPCI technology," said Ben Sharfi, chief executive officer of General Micro Systems Inc. (Rancho Cucamonga, Calif.), a maker of VME and CompactPCI boards.

An Intel spokesman said that the semiconductor giant bought Ziatech in 2000 during a communications acquisition binge that totaled more than $15 billion. Then its strategies for the segment changed.

"We decided to sell off the legacy products and focus on more forward-looking technologies," said the spokesman. "The older technology isn't central to our strategy."

But General Micro Systems' Sharfi believes Intel sold the business for so low a sum "because it basically has no value. They did well to get $3.8 million." Sharfi added that between 15 percent and 20 percent of his company's business comes from CompactPCI.

Sharfi and some others believe that CompactPCI may be relegated to a role as a legacy product as the telecommunications business shifts toward server blades, which would not use a shared CompactPCI bus across the backplane.

Specification blamed

Sharfi, who is an executive member of PICMG, blames PICMG's efforts to go to a 3.X specification. The proposed spec, he said, has disenchanted some customers who had invested heavily in PICMG 2.X.

"Several of our customers who went from VME to CompactPCI are shifting their focus back to VME again," Sharfi said. "The handwriting has been on the wall for a while. By the end of 2003, the CompactPCI market will finish at sub-$100 million. We're seeing it shrink down to nothing."

Several analysts said they had warned the industry about the questionable future of the market for CompactPCI merchant boards. "We've been forecasting the death of CompactPCI as a board business for years, and people haven't listened," said Jerry Krasner, vice president of Embedded Market Forecasters. "The CompactPCI merchant computer boards market is fading into oblivion."

Krasner contends that Intel stood to make very little revenue from the Ziatech business and could continue to sell its chips to board makers whether it owned Ziatech or not. "What Intel had was a white elephant," he said. "Sales had dropped dramatically, and they were giving away reference designs as a way of pushing silicon."

But Performance Technologies (Rochester, N.Y.) and others in the CompactPCI business argue vehemently that the sale price of Ziatech has no bearing on the future of the CompactPCI boards business.

In an outline of its strategies, Performance Technologies said that the acquisition more than doubles its total available market in the embedded-systems sector. It creates a complementary product line having no overlap with the company's existing products, and it opens up new markets for Performance in telecommunications, data communications, medical, transportation and military applications.

Analysts described that as a sensible strategy, especially if telecom moves toward server blades. "CompactPCI's growth will now depend on penetrating other markets more solidly," said Gulliksen of Venture Development.

Performance Technologies executives also argued that the telecommunications migration toward server-blade technology won't have a big effect on CompactPCI. PICMG's 3.X specifications, which are aimed at server blades, are likely to be employed in core-level applications, while CompactPCI will serve "at the edge" of the telecommunications infrastructure. As such, 3X technology would be used in large, high-bandwidth optical switches, while CompactPCI would serve in wireless basestation controllers, WAN switches and voice-over-Internet Protocol media gateways.

"We don't think that one is going to obsolete the other," said Ed Bizari, director of marketing and business planning for Performance Technologies.

Still viable

CompactPCI makers also said that their technology will continue to play a major role in the telecommunications industry over the next few years while customers wait for the first PICMG 3.X products to be unveiled.

"The industry is rolling out PICMG 2.16 products today," said Stuart Cohen, vice president of marketing and business development for RadiSys Corp. (Hillsboro, Ore.). "It will bring in the 3.0 products for test, evaluation and field trials over a period of time, which will then lead to production runs. But it's going to take time for that natural evolution to take place."

"You can't undercut an industry with something that doesn't exist yet," added Bizari of Performance Technologies.

Responding to questions from EE Times, Force Computers, one of the biggest sellers of CompactPCI technology, said the market is still growing. Roland Chochoiek, director of corporate marketing for Force GmbH, acknowledged that the technology no longer enjoys the three-digit percentage growth rate it had in the early days of adoption but still has double-digit percentage growth. "The architecture is being adopted in more and more application areas and market segments outside the telecom space," Chochoiek wrote. "CompactPCI is still on a growth curve."

Still, many analysts say that the selling price of the former Ziatech unit can't bode well for the future of CompactPCI. Moreover, the board market is still dependent on the health of the telecommunications industry, they say.

"If telecom were to bounce back today, CompactPCI would sell, and you might see it return to the $500 million or $600 million mark by 2006," Gulliksen said. "But telecom's not coming back today or next year. It might take several years."

If the recovery takes too long, Gulliksen added, the telecommunications market could jump right into server-blade technology. He declined to quantify his forecast for CompactPCI if that happens.

Whatever the economy, most observers agree that Performance Technologies' purchase of the Ziatech unit was a bargain at $3.8 million. "It's not as if Intel just jettisoned the trash," said Paul Zorfass, senior analyst for International Data Corp./FTI (Framingham, Mass.). "Ziatech didn't give Intel the profit and growth it wanted.

"But someone can still make money off this."