Aiming to bring down costs and reduce the need for incentives, NREL's engineers are studying mirrors made from lightweight polymers instead of glass and receiving tubes that will absorb more sunlight and lose less heat. They're also working on solar power's biggest problem: how to store some of the heat produced during daylight hours for release later on. "In the Southwest particularly, peak loads are in the daytime, but they don't end when the sun goes down," said Mark Mehos, an NREL program manager. People come home from work, turn on lights and air conditioners. Before long they may be plugging in electric cars.

Last year the first commercial solar plant with heat storage opened near Guadix, Spain, east of Granada. During the day, sunlight from a mirror field is used to heat molten salt. In the evening, as the salt cools, it gives back heat to make more steam. In Arizona the Solana Generating Station will also use molten salt for storage. When it goes on line in 2012, three square miles of parabolic troughs will produce 280 megawatts for Phoenix and Tucson. Solana is being built by a Spanish company, Abengoa Solar—an indication of just how far, in the development of this technology, the United States has fallen behind.

Back in the 1980s, an engineer named Roland Hulstrom calculated that if photovoltaic panels—the other big solar technology—covered just three-tenths of a percent of the United States, a 100-by-100-mile square, they could electrify the entire country.

People thought he wanted to pave the Mojave with silicon. "The environmentalists got up in arms and said, You can't just go out and cover a hundred miles square," Hulstrom said recently as he sat in his office at NREL. But that's not what he meant. "You can cover parking lots with photovoltaic. You can put it on house roofs."

Twenty years later, PV panels still contribute only a tiny amount to the nation's electricity supply. But on rooftops in California, Nevada, and other states with good sunshine and tax incentives, they're a sight almost as familiar as air conditioners—and though not yet as developed as solar thermal, they may have a brighter future.

Right now the panels are expensive, and they provide an efficiency of only about 10 to 20 percent, compared with the 24 percent of parabolic troughs. History more than physics is to blame. After the solar bust in the mid-1980s, many of the best engineers migrated to the computing industry, which uses the same raw material—silicon and other semiconductors. Following what is called Moore's law, microprocessors doubled in capability every couple of years, while solar languished. Now some of the engineering talent is moving back to solar.