Cypress' solar-power gambit bears fruit in SunPower IPO
Tuesday, January 17, 2006
When free-market champion T.J. Rodgers touts solar photovoltaics and invests in solar-cell specialist SunPower Corp., you suspect the days of government subsidies for residential solar electricity are numbered. When SunPower subsequently goes public and sustains a higher market cap than many semiconductor companies, it suggests that Rodgers, the CEO of Cypress Semiconductor Corp., timed the investment right.
Rodgers insists that solar cells manufactured on a semiconductor process will yield better returns on investment than most standard CMOS digital devices. He recently predicted that within five years, Cypress will be known as much for solar cells as for memory or communications processors.
But SunPower founder and CTO Dick Swanson is the first to admit that the journey to acceptance of solar power has not been a linear path to profitability for his company, which completed its IPO November last year. In fact, had Swanson and Rodgers not been Stanford University alumni chums, SunPower might not have recovered from a 2001 slump.
Swanson was an EE professor at Stanford in the 1980s when, under grants from the Department of Energy and the Electric Power Research Institute, he began working with silicon-based solar cells in a centralized solar-concentrator environment. At the time, utility and oil companies alike were still pursuing a circa-1970s vision for applying centralized-system economies of scale to solar power. Concentrators, used either for direct electrical conversion or for heating a turbine, would be located in central towers to give utilities control over the technology.
During the 1980s and 1990s, improvements were regularly made in the conversion efficiency of solar photovoltaic cells and in their cost per megawatt to manufacture and operate. Over the past decade, the concentrator concept was scaled down to emphasize systems sufficient to power a building. But the concentrator model found favor only in a few European demonstration projects and utility companies gradually lost interest in trying to exploit solar energy directly for electric-power production from a centralized grid.
SunPower continued to work in its own small fab on a cell design that used rear contacts on a piece of silicon, with the entire front surface usable as a solar collector. Luckily, the company was able to tap the same base wafers for development of such components as detectors and sensors.
"Around 1992, the component business was what kept money coming in the door," Swanson said. "Honda came to us around the same time to develop flat plates for a racing car, but that was one of those projects, like the later NASA plane, that were more about raising our profile than bringing money in the door."
"We knew that to develop a cell appropriate for residential solar panels on a cost-effective basis, we had to leave the world of photolithography and look at low-cost screen-printing techniques," Swanson said. "But 2001 was not an optimal time to raise money. When T.J. offered to make his own investment as a placeholder, we had been looking for venture investments."
When Rodgers met up with Swanson in 2001, SunPower was designing new solar cells for panels to be installed on rooftops, using N-type silicon float zones on a 5-inch2 cell. In his first attempt at interesting the Cypress board in an investment, Rodgers got nowhere, so he cut SunPower a personal check for $750,000 to keep the company going until he could sway his board. Cypress later invested $150 million in SunPower, beginning with an initial $8 million investment.
"The board originally said, 'Are you crazy? You're a communications company, you're doing PHY-layer stuff, 80Mbps, yada yada. You're going to lose your strategic focus,'" Rodgers said. "It took me 15 months to get them to invest."
Rodgers said the generation of 1W of power by solar means cost approximately $60 in the 1970s, compared with $2.99 today. The real economic advantage is realized in rural distribution systems, he said, where power generation can be taken down to $1.50 per watt. SunPower's 32-by-64 inch panels contain 72 solar cells generating 150W and sell for approximately $700.
After the Cypress investment, as SunPower began experimenting with new cells at Cypress' CMOS fab in Round Rock, Texas, Tom Werner, a former 3Com Corp. executive and the former president of another Cypress acquisition, Silicon Light Machines, was brought in as SunPower's CEO. Werner said that by mid-2003, as consumers were hit with the triple-whammy of the Iraq war, escalating oil prices and mounting evidence of global climate change, investors began to realize that the "reverse" model of solar-electricity generationwhere a residence or small business inserts energy into the gridmade economic sense.
"Once the decentralized notion of electricity generation becomes accepted, it's hard not to gain enthusiasm," Werner said. "The mainstream electrical market is a $1.2 trillion market, so if solar goes mainstream and you can grab just a chunk of that business, it's very substantial."
After proving its solar cell in the Round Rock pilot line, SunPower acquired a former disk-drive plant in the Philippines from Nippon Electric Corp. and converted it to a specialized silicon-wafer plant capable of producing 32 million wafers per year. The fab lines use continuous-flow processing and can use circuit-board methods, such as screen printing with squeegees, instead of photolithography. Photoresist alone would cost more than the selling price of a solar cell, Rodgers said.
"We take a 6-inch wafer, saw it down to 5-inch size and sell the finished cell for six bucks," Rodgers said. "This plant manufactures more wafers in a week than our other fabs manufacture in a quarter."
The 21 percent efficiency of current SunPower cells compares with typical photovoltaic-cell efficiencies in the industry of 14 percent or 15 percent, Rodgers said, and more improvement is possible.
Werner said that lab breakthroughs in polarization, where changes in the cell configuration and grounding polarity improve current-leakage specs, have not yet been implemented in production cells. "We foresee further improvements in conversion efficiency and we fully anticipate the Philippines fab being capable of turning out the equivalent of 100MW a year," he said.
Loring WirbelEE Times
This article was printed from EE Times - Asia located at: http://www.eetasia.com/ART_8800403658_765245_bfbc7c85_no.HTM
posted by philpower @ 8:08 AM,
