To make a real dent in global warming, a clean-energy technology has to be cost-competitive with coal - without long-term subsidies.

At least that's what Vinod Khosla argues. Founder of his eponymous Khosla Ventures and, more famously, a co-founder of Sun Microsystems, he says clean energy won't grow large enough to slow climate change if it isn't cost-effective..

It also must be capable of delivering power whenever utilities' customers want it, he says, and needs to be able to provide huge amounts of energy - enough to replace half the market, he says. "It needs to be 50 percent to be a climate-change solution, and it has to work in India and China," he says. "Without that, we're just making ourselves feel good."

How does this stack up against his own investments? The Silicon Valley venture capitalist is well-known for backing biofuels - and has also made investments in solar-photovoltaic (PV) companies, which convert sunlight into electricity. But he doesn't think those technologies can save the world. A good investment is different from a good climate-change solution, he says.

Of all the technologies he's backed, Khosla seems keenest on solar-thermal technology. It makes electricity using the sun's heat, instead of its light, and has the best chance of being cost-competive, he says. (Khosla Ventures invested in the solar-thermal startup Ausra.)

Other financiers also are once again attracted to the decades-old technology. In July, Beit Shemesh, Israel-based Solel Solar Systems announced it's building the world's largest solar park in California's Mojave Desert after San Francisco-based utility Pacific Gas and Electric Co. agreed to buy all 553 megawatts of the park's capacity. The same month, Spain's Acconia Energy closed $266 million in project financing for a 64-megawatt solar-thermal plant operating in Nevada. Projects are also underway in Spain, Morocco, Algeria and Egypt, among other places.

Why is this retro technology getting all this action? Solar-thermal is already far cheaper than PV, for one thing, and advocates believe the cost could drop drastically. According to the U.S. Department of Energy, solar-thermal projects delivered electricity for between $0.11 and $0.18 per kilowatt-hour in 2000, compared with between $0.24 and $0.30 per kilowatt-hour for PV. Of course, solar-PV competes with retail electricity prices, while solar-thermal projects would sell at wholesale prices to utilities.

But solar-thermal is far closer to being competitive, even though it's been installed in far smaller amounts than PV. A National Renewable Energy Laboratory study has forecast that solar-thermal plants should be cost-competitive with natural gas by 2020. And according to Terry Peterson, a solar power consultant at the Electric Power Research Institute in Palo Alto, Calif., there's only about 400 megawatts of it installed worldwide, compared with more than 5 gigawatts of PV.

It also turns out that storing heat is easier and cheaper than storing electricity. By heating water, oil or even rocks in a well-insulated tank, solar-thermal electricity can be stored as heat until utilities need it, or saved for a cold or rainy day when the sun isn't up for making more.

And, scientifically, it seems to make sense. According to the National Renewable Energy Laboratory, the potential solar-thermal-power capacity in the U.S. Southwest alone would be enough to generate four times the current annual electricity generated in the entire country. "The resource is almost unimaginably huge," Peterson says. "For the next 10 years, we should be building solar-thermal plants like crazy."

Of course, solar-thermal's history offers plenty to be skeptical about. The first plants were built in the 1970s and 1980s, and the technology went nowhere. Luz, a company that built two demonstration plants in California, went bankrupt. Some blame a lack of government backing after the oil crisis ended; others blame a nonexistent market and a paucity of funding for larger plants.

Because the plants must be large to be economical, and because the customers for the technology are utilities - notoriously conservative buyers - the technology has had a slow start even this second time around. While solar-PV has been booming for two years now, solar-thermal projects are just being built. They could again have trouble growing large enough to be cost-competitive.

Of course, VC backing, recent deals with utilities and a new pro-green climate could change the scene. The question is whether solar-thermal can realistically grow large enough fast enough to reverse climate change.

Khosla sees no reason it shouldn't make up at least 50 percent of all new capacity within three years. But Peterson demurs: Given the small base today, he says it's more likely to take decades to become a dominant technology.