SAN JOSE, Calif. The hopes and frustrations of the domestic flat-panel display industry can be glimpsed in this quiet corner of Silicon Valley, where Candescent Technologies Corp.'s manufacturing plant stands as an empty shell.
At the time ground was broken 16 months ago, with dignitaries including the mayor of San Jose in attendance, the plant was scheduled to begin producing flat-panel samples by early 2000, with volume shipments starting later this year. Instead, the 311,000-square-foot building has been mothballed, and production remains at least two years away.
Traditionally, the U.S. display industry has consisted of small, undercapitalized suppliers, surviving by serving niche markets too small for the giant Asian producers to bother with. But at its 1991 launch, Candescent claimed it would change all that, promising to crack the Asian monopoly by building a large-scale manufacturing infrastructure in the United States and delivering superior display technology.
Nine years later, the company has yet to sell a single display, though Candescent's executives insist they are making notable progress on perfecting their ThinCRT variety of field-emission display (FED). Nevertheless, skepticism lingers about the prospects for the domestic merchant suppliers in general and Candescent, originally named Silicon Video, in particular."None of them are very successful right now," said Barry Young, vice president of Display Search, an Austin, Texas, research firm.
Some might say the United States never had a realistic chance of becoming a major flat-panel producer, given the entrenched position of the Japanese and Korean manufacturers. Yet that argument is about to be demolished in Taiwan, where a mammoth crash-investment program will soon establish that country as one of the leading players in this $18 billion market. With half of the world's notebook computers already being produced on the island, Taiwan believes it has a built-in market that can support a half-dozen or so major domestic flat-panel manufacturers.
Although the U.S. display industry is hoping that emerging applications such as microdisplays and organic light-emitting displays will improve its fortunes, the mortality rate remains high. Over the last year, for example, Xerox and Guardian Industries both threw in the towel on their respective active-matrix LCD ventures, Micron Technology divested itself of its FED business and Agilent Technologies beat a quick retreat from the miniature-LCD relationship it had formed (as Hewlett-Packard) with Displaytech Inc. Both Candescent and PixTech, which is currently producing FEDs, replaced their chief executives in hopes of jump starting their prospects, and Motorola Inc., although issuing optimistic statements about its FED technology, is reluctant to make firm commitments on when it will go into full production.
From day one, Candescent declared it would develop a flat-panel display superior to the active-matrix (AM) LCD and aimed at the heart of the AM LCD market: notebook computer displays. It attracted the attention and dollars of computer makers such as Compaq and Hewlett-Packard, and of the Defense Advanced Research Projects Agency (Darpa), which was growing anxious about the lack of U.S. participation in what it viewed as a strategic industry.
Candescent's executives claimed they could produce a flat display comparable in brightness, contrast, viewing angle and video response time to cathode-ray tubes (CRTs), and Harry Marshall, the firm's gregarious CEO, managed to raise about $400 million for its quest.
According to the company's Web site, "The combination of high-voltage phosphors with a low-voltage switching cathode produces a ThinCRT that Candescent believes will be from 25 percent to 50 percent more power efficient than a comparable AM LCD." Power efficiency should be independent of display size, whereas AM-LCD power efficiency declines appreciably in small displays, it added.
Its displays would have many other advantages, the company claimed. By using fewer manufacturing steps, Candescent would achieve a cost advantage over AM LCDs, with lower yield losses at larger display sizes. Since its display wouldn't require a backlight, OEMs could use that space for other system electronics. Finally, since Candescent's displays wouldn't use liquid-crystal material, organic polarizers or backlights, they could be used in conditions of extreme cold and heat.
Based on those claims, Candescent raised $72 million from two offerings of preferred stock in 1995 and 1997, with blue-chip venture capital firms such as Sevin Rosen, Sierra Ventures and New Enterprise Associates participating. Hewlett-Packard, which was co-developing the FED technology with Candescent, invested $15 million in the second round. A sale of subordinated debentures in 1998 raised another $125 million. An initial public stock offering scheduled for late 1997 was shelved, however.
Candescent also received at least $40 million from government sources, including Darpa, according to estimates. Help came from local government, too. To lure a planned manufacturing plant to its city borders, Morgan Hill offered Candescent a $9 million loan, which would not have to be paid back so long as the plant remained in the city. But the company instead accepted a $10 million line of credit from San Jose, which Candescent could use to buy equipment for the plant. As part of that deal, the city agreed to relocate a road leading up to the planned facility, according to Leslie Little, an official with the San Jose Redevelopment Corp.
A potentially significant deal was also struck with Sony Corp., which agreed to spend $50 million and share its technical expertise with Candescent to help scale up and commercialize the FED technology.
A festive mood, then, attended the groundbreaking ceremonies for the manufacturing plant in November 1998. Mayor Susan Hammer helped turn over the first spade of dirt. The company arranged for a performance by a dance troupe it called the Thin Heads, who wore brightly colored costumes identical to those then in vogue in Intel commercials, with the exception of masks that resembled flat-panel displays. At the end of a dance, one performer whipped off his mask to reveal a smiling Harry Marshall, CEO.
Four months later, the smiles were gone and so was Marshall. He was removed as CEO, although he was kept on as chairman of the board. Roughly 10 percent of the company's employees were laid off, and the manufacturing startup date was postponed. It had become apparent to Candescent's board that the FED technology wasn't ready to be commercialized, and a huge amount of cash still had to be raised to complete the manufacturing plant.
"The anticipated improvements were not [coming] quite as fast as they had hoped," said board member Bernie Vonderschmitt. "I think Harry [Marshall] did a very good job in the early development phase, but we thought we needed somebody a little bit more, shall we say, focused in making sure goals are set on time and in monitoring once a quarter to make sure you stay on the beam."
The delay in perfecting the FED technology wasn't unique to Candescent. "Our conservative forecasts for FED turned out to be wildly excessive," said David Mentley, senior vice president of Stanford Resources (San Jose). "There's been a lot of overselling and oversimplification of the challenges. It's been a tougher technology development problem than anyone wanted to admit in the beginning."
One of the toughest hurdles to clear was related to the use of high-voltage phosphors, said Stewart Hough, Candescent's marketing manager. To prevent arcing, Candescent has had to create a wall structure, designing spacers that could be placed a centimeter apart all over the display, yet remain invisible to the viewer. "Making them invisible is not trivial," but the company has managed to overcome the problem, he said.
Also time-consuming has been the effort to scale the switching voltage from 50 V down to 10 V. "That will give you a 20-times reduction in power consumption," Hough said. "That is where we have spent most of our efforts."
Work has also continued on reducing luminance degradation over time, measured as decrease in brightness with use. Here, Candescent has managed to replicate the performance of existing CRTs and TVs, Hough said.
On the issue of reliability, measured as mean time between failures, the company has a goal of 50,000 hours. "Reliability analysis and design has determined that this goal is achievable," and testing is ongoing to verify the numbers, he added.
Candescent has also has been slowed because of its collaboration with Lawrence Livermore Laboratories to eliminate the use of steppers in building its displays. Steppers are expensive and slow, Hough said. By going to an ion-implantation process, Candescent can bring down the costs of its displays significantly. The work with Lawrence Livermore has been very promising, he said. "This has taken a lot of time and a lot of financial resources, but the potential payoff is worth it."
The 5.3-inch prototype FED shown by Candescent is a vibrant color display offering a wide viewing angle. The company more recently has begun showing a 13.8-inch color display, which it claims proves the technology can be scaled successfully. But can either one be built in high volumes, and at a price the market will accept? "We believe we'll have a cost advantage over AM LCDs," said Kyle Baker, senior director of marketing. "That analysis has been gone over line by line with industry analysts and people in the display industry."
But not all agree. "The selling job was that this was a lower-cost solution with fewer steps than LCDs. We never subscribed," said Stanford's Mentley. "It's just about the same complexity as LCDs. There are so many issues-vacuum integrity problems, lifetime problems, electron optics issues. It's not something one or two companies can easily solve with tens of million of dollars. It takes hundreds of millions of dollars. That was the case for AM LCDs and CRTs."
What's more, the dominant LCD industry isn't standing still, said Young of Display Search. "They're a juggernaut, 25 companies working on problems associated with cost reduction." Those companies are also busy improving the features of their displays, which could make it harder for FEDs to crack mainstream applications. "I think the notebook market is moving away from them," Young said.
Hough acknowledged that competitive threat: "The LCD industry has a hundred engineers working on problems for every engineer working on FEDs."
Candescent's ambitions have never been small. The company's strategy is to build multiple manufacturing plants to address multiple markets. Though it expects to begin with small displays, it foresees its flat panels being used in applications ranging from autos and notebook and desktop computers, to wide-screen televisions, in form factors running from just a few inches to 30 inches and larger.
"We believe that the approach of entering into these large markets offers a lot of advantages, including better management of intellectual property," said Baker. "And we don't want to be a $30 million or $50 million company, we want to be a half-billion-dollar-a billion-dollar-company."
But that's going to take some big bucks, and Candescent hasn't announced any major investments in more than a year.
"They're in trouble, clearly," said Chuck McLaughlin, an analyst who covers the display industry for McLaughlin Consulting Group (Menlo Park, Calif.). "It gets harder and harder to get incremental money. You hate to be the guy who puts in the last hundred million." Venture capital funds may be reluctant to invest in a capital-intensive company like Candescent, given the riches lately being made with Internet companies.
Candescent is in talks with a number of potential manufacturing partners that could invest in the company, said David White, president and CEO. And capital markets worldwide remain to be tapped. "We've used those markets successfully in the past, and I think they'll be an important part of the strategy." White said he's optimistic a deal will be concluded soon.
The company will need to raise about $400 million to get its manufacturing plant into full production, White said. He estimated production there could start in about two years, given the lead times for receiving and installing capital equipment.
"We've been fairly quiet the last year, putting our nose to the grindstone, nurturing and developing our relationship with Sony," he said. "The Sony collaboration has gone exceptionally well. We've met every milestone of the program."
The money can be found to build the company, Vonderschmitt believes. "Look at all the Internet companies being formed. That tells me there's an enormous amount of money available," he said. An IPO also remains a possibility within a year, he added.
As for the technology, "The jury isn't in, but I don't think there are any significant barriers. I think they've gotten to the point where the technology is pretty well-defined," Vonderschmitt said, "but it's not anywhere near the mature point where you can predict all the yields and variables."