Billat, who analyzes capital equipment companies, brought some sobering words to the test-engineering audience here Tuesday (Oct. 30) about the way Wall Street evaluates its part of the electronics industry. "The ATE automatic test equipment companies' financial performance is viewed as an early indicator of trends, but with fewer technology drivers, and with ATE customers considered as a necessary evil in the semiconductor food chain," said Billat.

She claimed that the business model for the ATE industry is broken. "When five years ago a piece of ATE cost $1 million, two years ago it went for $2 million and today you have to pay $3 million for it, there is a need for an overhaul."

This perception, among other factors, has catapulted the top three ATE companies, Agilent, Advantest and Teradyne, to the bottom of market evaluator VLSI Research's list of the top 15 semiconductor equipment manufacturers. "Customers can't afford these behemoths any more," Billat said.

"The time-to-volume production cycle needs to be accelerated," said Billat, but she wondered if the industry can accomplish that. "Designers are still being rewarded for time-to-tapeout, while volume production is profitable only in the first six to nine months, so you need to ramp volume production as soon as possible."

The cost of test is running counter to the Moore's Law and functional test still represents a vast majority of test, said Billat. "Meanwhile, we are hearing that the design community is very comfortable about producing chips at the 70-nanometer level soon. How do you test a billion transistors?" She called for a push to structural testing as an antidote to managing the complexity of dense next-generation chips, especially system-chips.

Billat warned the test community that when test costs equal 2 percent or more of a chip's entire cost, it might cause the ripple effect that the photomask industry is seeing.

"When the costs for masks for photolithographic processing became astronomical, chip makers found ways to start using less masks, forcing a consolidation of that industry," said Billat. "The same could happen for test companies if costs cannot be contained."

Besides the obvious moves toward design-for-test and strip testing as alternatives to today's singular testing, Billat suggested that the test industry form an association on par with what the process fabrication industry has done. "Since the Fabless Semiconductor Association was founded in 1984, it has been able to help fabless companies balance their business plans with industry supply-and-demand fluctuations," she said. "Something along these lines might help test companies forecast equipment demand in up and down cycles and avoid the consequences of double and triple ordering."

Billat reminded her audience that Wall Street picks winners and that among EDA, front-end processing equipment and back-end processing companies, the valuation is lowest for the back-end vendors, a category that includes assembly and test: 6, 4, and 3.5 respectively.

"The link between design and test is not obvious to Wall Street, but they will pay a premium for what they do understand," said Billat. To get noticed by investors, she called for the test industry to get the word out on how design equipment is being linked to test equipment.