(FORTUNE Magazine) – One morning early in August, a fire truck, an ambulance, and a convoy of vehicles crammed with machinery, spare parts, and test instruments arrived at Willow Springs Race Track in the desert outside Los Angeles. They were to test a hybrid power train for automobiles unlike any in production in the world. For Ben and Harold Rosen, who were leading the effort and financing it out of their own pockets, Willow Springs was supposed to provide the first solid evidence that their bold and expensive gamble might succeed.

The test was a dismal failure. During 14 hours at the track, the Saturn car that housed the new power plant didn't budge. First, an improperly installed fuel switch cut off gas to the main engine. Then another malfunction kept the engine from revving up to speed. Problem piled upon problem; finally, near sunset, the crew gave up. They'd shelled out thousands of dollars to rent the track when they could have accomplished just as little in a vacant parking lot.

Such is the struggle to develop a propulsion system that will take autos into the 21st century. Although the internal combustion engine developed by Gottlieb Daimler has reigned supreme for 111 years, research is accelerating on alternatives that will use much less gas and emit almost no pollution. Activity is especially brisk in Europe and Japan, where gas costs up to $4 a gallon and the skies are dirty brown. In the U.S., battery-powered electric cars by General Motors and Honda will arrive on the market starting later this year. And GM is working with Ford and Chrysler in a federally funded program to develop all-new vehicles early in the next decade.

Next to these giants, the Rosens are mites. They have just 60 employees and get no government grants. To date, they have spent about $10 million, nearly all of it Ben's. But these guys aren't your typical amateurs. Harold Rosen is a gifted inventor, a former Hughes aerospace executive who 33 years ago pioneered the geostationary satellite, a profound innovation that makes possible today's global telephone and television networks. His younger and more famous brother, Ben, has built a fortune of nearly $100 million by starting technology companies, including Compaq Computer.

What the Rosens are developing is a propulsion system that is, in theory, more powerful, more fuel-efficient, and less polluting than internal combustion. It employs two power sources--a small turbogenerator and a carbon-fiber flywheel--to run electric motors that drive the wheels of a car. The Rosens want to use this power train to launch the most ambitious independent carmaking effort since steel magnate Henry J. Kaiser set out to build Jeeps and small autos after World War II. They call their company Rosen Motors.

If the internal combustion engine has served us for more than a century, why change it now? Antipollution laws, for one thing. California is demanding that 10% of the cars sold there by major manufacturers by 2003 produce no pollution; other states are likely to follow suit. Another motivation: Oil is a finite resource that will become increasingly expensive. And finally, technologies to build better engines are now available. A familiar example: the electronic control circuits that make ordinary engines more efficient.

Researchers are looking for solutions beyond the clunky battery-powered electric cars that are reaching the market. Despite years in development, batteries still can't store enough energy, limiting the range, speed, and drivability of electric cars. So automakers are focusing on hybrid systems like the Rosens'.

Hybrids try to capitalize on the fact that a car engine must do two quite dissimilar things: accelerate and cruise. An engine that can accelerate a 3,500-pound vehicle from a dead stop generates much more power--and burns far more gas--than is needed to keep the car rolling at constant speed. In a hybrid, one engine is optimized for acceleration, another for cruising.

The Rosens' system incorporates a flywheel for acceleration and a turbogenerator for cruising. Both produce electricity that in turn powers so-called traction motors that drive the wheels. A sophisticated electronic controller manages the system. The turbogenerator is the only component using fuel: It burns small amounts of unleaded gasoline and produces virtually zero emissions.

Since there is no market for clunkers, the Rosens have set rather audacious performance goals for the new power train, aiming to beat Detroit's federally funded partnership. GM, Ford, and Chrysler have promised to develop a family-size car that gets up to 80 miles per gallon of fuel, vs. 27 today, and produces far fewer pollutants; they plan to deliver working prototypes by 2003.

The Rosens are aiming for a much faster car--one that goes from zero to 60 miles per hour in six seconds, vs. 12 seconds for today's average car--that would also emit virtually no pollutants. For fuel economy, they want to hit between 40 mpg and 80 mpg, depending on the size of the car and its drag characteristics. Their timetable is far more aggressive than Detroit's. They hope to demonstrate a car capable of reaching the goals next year, and to put their power train in a production vehicle by 2000. Jokes Ben: "We're trying to prove that a dog can talk. A year from now, it will be saying the Gettysburg Address."

But it turns out that satellites and PCs were a snap compared with the complexities of motor vehicles. Says Harold: "Developing this is like peeling an onion. It has been more difficult than our first communications satellite. But every development program has similar characteristics: one damn problem after the other. The very first satellite we built blew up when it reached orbit. You have to persevere."

Pulling this off would be difficult even for an established automaker. Experimental engines such as the Stirling, Orbital, and two-stroke have died aborning, despite support from industry giants, and the Wankel rotary pioneered by Mazda seems to be fading from production. David Cole, the director of the University of Michigan's Office for the Study of Automotive Transportation, gets to peek at many new technologies. Of the Rosens' project he says, "It is all potentially doable. The real question is manufacturability and cost. It is a long shot, it is really a long shot."

The Rosens don't disagree. "This is highly ambitious and highly unreasonable," Ben says over lunch at the Sky Club atop Manhattan's Met Life Building. "You could call it Rosen's Folly. But Harold and I think it is the right solution. It is nothing you should try unless you can fail with impunity. We can do that, though it will hurt our ego, and our pocketbooks."

Failure would be new to the Rosens. They were born into a strict Jewish household in New Orleans. Their parents divorced, and their father, a dentist, moved out of the house when "Benji" was 8. Harold, a worldly 15, acted as a surrogate father to his brother. After serving in the Navy in World War II, Harold became an electrical engineer; later, so did Ben. When Harold became established as an engineer at Raytheon in the Fifties, he hired Ben, just out of college, to work under him.

Harold spent an entire career in electronics, putting in 37 years at Hughes in Southern California. He was awarded more than 50 patents and won a slew of engineering prizes. Alois Wittmann, a Hughes co-worker, says, "Harold is a realistic, practical guy who always comes up with a solution. I do not know of an unsuccessful project Harold worked on." Lean and wiry, Harold has been putting in 70-hour weeks to get the power train ready. He seems to have the vitality for it: He recently finished second in a company athletic contest by doing 42 pushups.

Ben left electronics to become a highly successful security analyst. He played a key role in the early history of personal computing: An annual conference he created was the forum where heads of fledgling companies met, made deals, and learned to think of themselves as an industry. With his natty wardrobe, acerbic wit, and fondness for clowning by balancing large computer cartons on his nose, Rosen was as much a persona as Bill Gates or Steve Jobs. He left Wall Street in 1982 to create a venture capital company, Sevin Rosen Partners. In addition to Compaq, where Ben is chairman, Sevin Rosen made canny investments in such groundbreaking ventures as Silicon Graphics, Cypress Semiconductor, and Lotus Development.

Although the brothers have lived on different coasts and worked in different worlds for most of their lives, they still finish each other's sentences. A rumpled, somewhat shy intellectual, Harold likes to twit his more urbane brother when he thinks Ben is putting on airs. Ben, meanwhile, describes Harold as "the smart one in the family" and defers to his technical judgment.

Their automobile adventure began 3 1/2 years ago, as Harold was winding up his career in communications satellites. Colleagues were working on electric-powered vehicles for GM, which had bought Hughes in 1986. Harold got a ride in an early test version for which Hughes had built a controller. He quickly decided the car was seriously limited by its batteries.

Curiosity aroused, Harold began looking for an energy system that would be emission-free like batteries but would produce more power. Soon he zeroed in on hybrids. A friend at NASA passed along a paper by an engineer who described a proposed high-speed turbogenerator that could be used in a hybrid. Recalls Harold: "It had on paper everything I was looking for. It had a catalytic combuster so there was essentially zero pollution, and a very high rotation speed, which meant it would work with a relatively compact electric generator."

Harold went to visit the author, Robin Mackay, who had co-founded a company, NoMac Energy Systems, in nearby Tarzana. There Rosen received "a rip-roaring presentation that got me all excited." He took the turbogenerator idea to Hughes's automotive section, but with GM already committed to batteries, Hughes decided not to pursue the idea. So, says Harold, he did the only logical thing: "I called my brother. He was looking for something to do."

Already something of a car nut (he owns a Porsche 928 and a $130,000 Mercedes-Benz SL600), Ben was phasing out of his venture capital firm. While still involved at Compaq and on the boards of Caltech, Columbia University's business school, and the Memorial Sloan-Kettering Cancer Center, he had joined five golf clubs and worked his handicap down to 18. When Harold called to propose starting a company together, Ben didn't hesitate.

Once Ben got hooked, he persuaded his partners at Sevin Rosen to join with him and Harold to buy the turbogenerator company. Two other venture capital firms contributed as well; recently Microsoft billionaire Paul Allen became an investor. To date $25 million to $30 million has been invested, and the Rosens say the company has a value of $50 million. They installed a new CEO, another Hughes retiree, and renamed the business Capstone Turbine.

Auto companies have been experimenting with turbines for more than 30 years. Modern turbines are powerful and fuel-efficient but also big and noisy, which is why they are used mainly in aircraft. The Capstone turbogenerator, by contrast, is just 38 inches long and 18 inches in diameter--about the size of a large picnic cooler. It weighs just 165 pounds and makes much less noise than a gasoline engine. Its sole moving part is a rotor shaft with a small gas turbine on one end and an electric generator on the other.

Though further along in development than any other piece of the Rosen power train, the turbogenerator still has a way to go. The initial version produces 24 kilowatts of electricity, and the Rosens are working on one that generates 45 kw for larger cars. Progress, says Ben, has been frustratingly slow: "When we went in, we thought, 'We can compress the time scale and do it quicker.' In fact, we can't. Capstone couldn't be more different from, say, an Internet investment, which is low capital and early gratification. This is big capital and delayed gratification." He expects Capstone to break even by 1999.

Capstone had been working on so-called stationary applications for its turbogenerator--a single unit can provide backup power for an 11,000-square-foot office building, for instance. But Ben was impatient to move ahead with a car. Says Harold dryly: "My brother didn't want to wait, so we formed Rosen Motors to push the development of the complete power train."

Rosen Motors was incorporated in May 1993, with Ben as chairman and Harold as president and CEO. They enlisted yet another retired Hughes executive, William Grayer, to run day-to-day operations. A sign posted on the door of a lone office at Capstone designated it "World Headquarters of Rosen Motors." (The auto company now has its own machine shops, laboratories, and garages in an industrial park in nearby Woodland Hills.)

To complement the turbogenerator, Harold had to find another device to provide power for quick starts. He quickly settled on flywheels. Low-tech flywheels have been used for years on automobile crankshafts to help engines run smoothly. But researchers were increasingly investigating them as power sources. The idea is simple: A rotating object such as a potter's wheel possesses kinetic energy that can be converted to electricity. A flywheel can store excess energy from a turbogenerator and also save energy that would otherwise be dissipated when a car brakes.

The Rosens have set out to build a flywheel like none the world has ever seen. They want one that will pack enough power to let a driver peel out at a stoplight and that will be able to go on spinning for at least 1,000 hours--six weeks--when the car is parked. That way, even if a Rosen-powered car sits for weeks at the airport, say, no battery or starter motor will be needed to get it rolling.

Since no such flywheel is available commercially, the Rosens are building their own. Some 4.5 miles of carbon-fiber strips, each with 12,000 strands one-twentieth the width of a human hair, are soaked in resin and installed on a titanium hub. The finished flywheel has a diameter of about 15 inches and weighs 21 pounds. To minimize friction, it spins in a vacuum. It is mounted vertically on a shaft with an electric motor-generator, and is housed in a steel sphere.

Flywheels present sticky development issues, especially in the unforgiving environment of an automobile, where components are subject to bumps, turns, and collisions, not to mention extremes of heat and cold. When the wheel is spinning full speed at 55,000 rpm, its outer edge is moving at more than 2,000 miles per hour and is subject to centrifugal force of more than 500,000 times gravity. If it fails, it can disintegrate, flinging shrapnel, or spin off its axis like a buzz saw run amuck. A worker at a lab in Germany was killed last year while testing a flywheel rotor for BMW. For now, the Rosens are using a 270-pound canister made of steel and kevlar to contain the flywheel, but they hope to reduce its weight to a tenth of that (see box). Half the effort at Rosen Motors has been devoted to the flywheel. Says Ben wryly: "There is a good reason why there are no working automotive flywheels."

Another difficult task has been to integrate the turbogenerator and flywheel with an electronic controller and fit them in a car. Early on, Harold donated an almost new 1993 Saturn coupe as a test vehicle. "I was naive," he says now. The turbogenerator took up the entire engine compartment, while the flywheel and controller filled the trunk and intruded on the back seat. There was no way to shrink the system quickly enough to make a presentable compact car.

Harold solved the problem by leasing a $44,000 E320 Mercedes-Benz sedan. He and his engineers removed the sophisticated German power train and installed Rosen Motors innards. The Mercedes' trunk is so voluminous that it will accept both flywheel and controller, with room underneath for traction motors to drive the rear wheels. Since replacing the engine and transmission with an experimental hybrid power train was not exactly covered in the lease agreement, Rosen Motors eventually bought the Benz.

Eventually, all the exotic machinery should be invisible to the driver who slips behind the wheel of a Rosen-powered car. The flywheel will already be spinning silently at immense speed in the trunk. To start, he'll turn an ordinary key; instead of engaging a starter, it will send current via a central circuit, or bus, to the turbogenerator under the hood. The turbine will build to a muffled whine, which will stay constant as long as the key is on.

Next, shifting a lever to "Drive," the driver tromps on the pedal, sending power to the wheels. The system sucks current from both flywheel and turbogenerator to rocket the car to the speed limit and beyond, quietly and cleanly.

At cruising speeds, power comes from the turbogenerator alone; the flywheel provides surges of power to accelerate or climb a hill. When the car is coasting or coming to a stop, the flywheel takes up and stores electricity from the so-called regenerative brakes. A single 12-volt battery on board is used only for accessories like headlights.

Even if the Rosens succeed in developing a car that hits their performance targets, they will be only partway home. They will have to prove that the power train can be built in quantity and at low cost, survive durability and reliability tests, and meet environmental standards. After that, they'll have to find somebody to buy it.

The Rosens briefly considered becoming a full-blown car company, a la Tucker or DeLorean, and talked to designers about creating a body and chassis. But they found the capital costs would be onerous and the learning curve steep; at the same time, there was little to be gained by presenting the world with yet another car body.

Instead the Rosens decided to specialize in building power trains. Ben based the business model on Intel, which prospers by putting chips inside other manufacturers' computers. Just as many PCs today sport the logo "Intel Inside," Rosen envisages fleets of cars wearing the badge "Rosen powered."

It is a long shot indeed. The methods used to build internal combustion engines have been refined to generate enormous efficiencies and economies of scale. Says Ben: "My biggest worry is economic. We're going to have to tool, not for a hundred units or a thousand, but for tens of thousands, just to get a reasonable cost. That's going to involve a bet that the market will be there."

Harold is convinced that he and Ben are on the right course. Today some 40 automakers around the world compete with power trains that are essentially identical. The Rosen hybrid could give someone a distinct technological advantage. That will probably be a European automaker, given the Europeans' exposure to high gas prices and more environmentally sensitive customers. Adds Harold: "It is extremely likely that starting early in the next century, there will be a transition from internal combustion engines to hybrid electric vehicles. Our future is up to us. We have no excuses. If we do what we've set out to do, we'll have a winner. If we stumble, somebody else will make it happen."

Ben has funded Rosen Motors through 1997, by which time he will have burned through $20 million. He is unperturbed. "By then," he says, "we should pretty much know whether we'll have customers and whether we will achieve our goals." If it looks as though the company can continue, he intends to seek conventional venture financing and perhaps go public. At his age, however, making a contribution to society is higher on the agenda than making more money. As for brother Harold, Ben says, "Rosen Motors is a chance for him to change the world--again."

As August ended, the Rosens booked another day at Willow Springs. This time, though, they decided to test the Saturn thoroughly the night before. At the plant, the car's power train ran for three hours flawlessly. Then a tiny malfunction started a cascade of events ending in disaster. An electronic glitch caused a magnetic bearing supporting the flywheel to exceed its programmed limits. Mechanical bearings, installed as a backup, failed too. The flywheel spun off its axis and crashed.

Thanks to the containment vessel, no pieces escaped and no one was hurt. But the project was set back while technicians hunted for the bug. Harold persuaded Ben to postpone indefinitely a planned public debut of Rosen Motors until they were completely satisfied with their device. Both Rosens say they're disappointed, but they haven't given up their goal. Harold is philosophical: "Like I said, developing something new is one damn problem after another."