(FORTUNE Magazine) – When was the last time you heard about a bold CEO who has bet the company on some big, hairy project that may never pay off? Has the age of compliance turned executives into wimps with a calculator in one pocket and a user's guide to Sarbanes-Oxley in the other?

Sometimes it does seem that risk has gone the way of Bell Labs and the three-martini lunch. That's why consumer products companies release a bewildering array of brand extensions--Crest Whitening Expressions Refreshing Vanilla Mint toothpaste, anyone?--and movie studios crank out sequels and prequels and remakes. CEOs with cash to burn sleep more easily when they use it to buy back stock. The age of the corporate buccaneer is gone.

Look a little deeper, though, and it becomes apparent that when competition, stagnation, or investor pressure up the ante, 21st-century executives are just as capable of laying down a high-stakes gamble as any 19th-century robber baron. After losing ground to Airbus through the 1990s, for example, Boeing launched a revolutionary new airliner, the 787, made with reinforced plastic, like the stuff used in Formula One cars. With margins vanishing in its traditional hardware business, IBM reorganized itself as a service company using Internet technologies--spending a cool $10 billion over ten years to make the transition.

Jeff Immelt, the CEO of General Electric, lives with decisions made three CEOs ago, when Fred Borch poured money into computers, nuclear power, and jet engines in the late 1960s. The first two were stinkers--but the third has given Immelt the lift to place bets that could pay off for his successors. Each investment involves a large dollop of uncertainty, which Wall Street hates. But then, Wall Street hated the 707 and the 747--until it loved them.

For some companies, gambling is a way of life. Driven by the imperatives of Moore's law, Intel is spending nearly $5 billion this year to make chips with more circuits--and it will do the same in 2007, 2009, and 2011. Other companies make big bets once in a lifetime. Verizon and SBC Communications, for example, are investing billions to replace old copper-wire networks with fiber optics.

Innovation, says Immelt, is the "central necessity" of modern business. What follows are accounts of how five companies are pursuing that necessity.

Fuel cells

Take any new technology, goes an inside joke among Detroit's car guys, and General Motors has probably been running it on its test track for years. The problem has been getting it on the road. GM has a rich history of innovation. The company invented the self-cleaning exhaust pipe and has recently done well with innovative onboard telematic gadgets like OnStar. But too often it has found itself lapped--or worse, stalled at the start. For example, GM blew $1 billion on battery-powered cars in the 1990s before it discovered--surprise!--that no one wanted to buy a car they couldn't drive to work and back without recharging. Then there was the hybrid disaster. GM had worked on those gasoline/electric cars for years, then shifted its emphasis in the late 1990s to concentrate on hydrogen-powered fuel cells--only to be embarrassed when Toyota reaped a public relations and sales bonanza with the hybrid Prius.

Now GM is betting big on cars powered by fuel cells. The technical and economic challenges involved in stripping electrons from hydrogen atoms to create electricity and leave only water in the exhaust are immense. Moreover, competitors like DaimlerChrysler and Toyota are also pouring money into the pipeline. So are governments. The recent U.S. energy bill, for example, allocated $4 billion for research into hydrogen. It is impossible to say who is ahead. No question, though: This time GM is in for the long haul. It has already spent $1 billion on the project and anticipates spending another $1 billion by 2010, when it hopes to have designed a fuel-cell-powered car that is affordable when produced at high volumes.

GM created its first fuel-cell vehicle in 1968 and continued to dabble with the technology into the 1980s before backing off when progress slowed. In the mid-1990s, Harry Pearce, then vice chairman, gave the go-ahead to start again when independent researchers began to report breakthroughs. Larry Burns, 53, who holds a doctorate in civil engineering, is GM's vice president of R&D--and the man whose job it is to make fuel cells work. One of his first acts when he got the job in 1998 was to drive a fuel-cell-powered Opel minivan. He did not enjoy the experience, describing it as hot, noisy, and impractical. Still, Burns is a believer: "We felt that we had a chance--with hydrogen and fuel cells, electronic controls, and software--to radically change the automobile."

Under Burns, GM has been spending up to $150 million a year on fuel-cell research, a figure that is growing briskly as the researchers get excited. Burns says GM has been able to raise the amount of power stored in the fuel cell by 700% in the past six years, as well as increase its durability, reliability, and ability to start in cold weather. Its latest model, a five-passenger SUV called the Sequel, will go 300 miles between hydrogen fill-ups, and do it quietly and smoothly.

GM CEO Rick Wagoner calls the fuel cell "the biggest thing to hit the automobile in the last 100 years." It is the single largest item in the car giant's R&D budget. Far more than money is at stake in this project, however. If GM falters, it will mean a major setback in the race to reinvent the automobile--and reinforce the idea that GM has lost the innovative edge that once made it great.

Kimberlite diamonds

Finding buried treasure is a childish fantasy, usually reserved for the very young or the wildly impractical. But a team of mining professionals and diamond experts believe they're about to strike it rich in the wilds of western Canada. They may be sitting on the largest diamond mine in the world, with potential deposits of 70 million carats. That's equal to half of all the diamonds produced anywhere in 2003. On the other hand, the consortium might be digging a very big, very expensive hole. Finding out will cost them roughly $100 million over the next few years alone.

Kimberlites--pipes of volcanic rock where diamonds were formed 100 million years ago--are prevalent in African countries like Botswana, South Africa, and the Congo. But nobody seriously hunted for them in Saskatchewan until the late 1980s, when geologists prospected the Fort à la Corne region, about 100 miles south of Saskatoon. The results were intriguing enough that a joint venture between De Beers Canada and Kensington Resources, a Canadian mining company, decided to dig more deeply. By the late 1990s the partners had identified 9.5 billion tons of potentially diamond-bearing kimberlite spread over 85 square miles.

Looking for diamonds in kimberlite is like searching for a needle in a haystack using a pick and shovel. The tops of the kimberlites, some 100 feet across, generally lie several hundred feet below the earth's surface. So where are the diamonds? Inside, sprinkled randomly. Finding them therefore requires sifting through enormous quantities of dirt and rock: Diamonds may be glamorous, but finding them is not. For every 100 tons of material, the venture expects to find 12 carats of diamonds, worth about $2,000 wholesale. Some 70% of the diamonds are expected to be good enough to sell as gems. The largest stone found to date in preliminary test borings weighs 10.5 carats and could be worth $10,000 in its raw form.

To ensure that they dig in the right places, the prospectors have laboriously drilled 349 holes in the kimberlite, ranging in diameter from several inches to three feet. The sheer volume of material is enormous. Two of the kimberlites are the largest in the world, containing 50 million tons of matter. The joint venture controls 63 similar deposits.

The partners are spending about $30 million a year on advanced exploration and feasibility studies, and they are moving ahead on environmental and other approvals. Still, they won't make a decision on whether to go on digging until 2008, and receiving permits and constructing the mine could delay the start of actual operations until 2013. Construction of the mine alone will cost $1 billion.

Despite the long lead times, the partners believe they are playing into a rising market. Spending on diamonds has been slack for nearly two decades and has lagged well behind that of other luxury goods, but growing incomes in India and China, as well as the successful marketing of new kinds of jewelry, are expected to boost demand during the next several years. Since De Beers largely controls the market, it can introduce new supply as it chooses and thus has little fear of pushing down prices. Last year, for example, demand surpassed supply, driving diamond prices up 20%.

If prices (or fashions) falter, however, or if Fort à la Corne proves less rich than initially thought, the partners will be, quite literally, in a hole of their own making.

Joint Strike Fighter

In the defense industry, one contract can determine a company's outlook for decades. And if that contract also happens to be the most expensive in Pentagon history, well, the stakes just get that much higher. At Lockheed Martin, all eyes are on the F-35 Joint Strike Fighter, an aircraft on which the U.S. military plans to spend $220 billion over the next decade.

The fighter's biggest selling point is its affordability: At $44.8 million for the basic version, it costs 60% less than the F/A-22, Lockheed's most advanced jet. The U.S. and Britain have already signed on for 2,593. Other countries may buy as many as 3,000, according to Richard Aboulafia, an aerospace analyst with Teal Group in northern Virginia. Lockheed is betting that it can keep the JSF's costs down at every phase of the plane's life and still deliver a superior product. At stake is dominance in the export market. "We have the potential to be the airplane that's in everybody's air force," said Tom Burbage, head of the JSF program. But should the cost per plane creep up, potential buyers will balk and cut back or cancel orders, pushing the price of each jet higher and thus reducing demand further. Lockheed would then be looking at a much smaller pile of chips.

The project began in 1993 when the Pentagon started to solicit designs for a plane that could replace its aging F-16 fleet. It had to be able to evade radar; be accepted by the Marines, Air Force, and Navy; and be able to land three different ways: vertically, onto a runway, and onto a carrier.

After McDonnell Douglas dropped out, a bare-knuckles brawl ensued between Boeing and Lockheed Martin. In October 2001, Lockheed won, effectively pushing Boeing out of the tactical-fighter business. There are now 8,000 people working on the JSF, including partners Northrop Grumman and BAE. The first F-35 is expected to fly in 2006.

In one corner of a mile-long factory in Fort Worth, engineers are peering inside a molded hulk of metal --the cockpit of the JSF--atop a bright-blue platform. At the height of production for the F-16--the most successful military aircraft in history--the Fort Worth factory turned out a plane a day. Lockheed managers, their fingers crossed, say they expect demand for the JSF to be just as brisk. When will they know if they're on to a winner? "Seven to eight years," says Aboulafia. In the defense business, technical prowess counts--but so does patience.

Siberian oil

Drilling for oil is one of the scariest, costliest, hold-your-breath-and-go-for-it gambles in all business. Oil companies have no choice: They must replace the crude they pump out of the ground or go slowly out of business. Outside of places like Saudi Arabia, where the sands fairly drip with oil, the easy stuff is long gone. Producers therefore must take ever greater risks--financial, logistical, market, political--to keep a steady flow. Consider Russia. With reserves estimated at 171 billion barrels--a fifth of world supply--it is the largest oil producer outside the Middle East. Its deposits in eastern Siberia are especially attractive because they are close to the fast-growing Asian markets. But investing in Russia is a high-stakes crapshoot--in the case of Exxon Mobil, to the tune of some $4 billion.

The U.S. oil giant signed its first deal in Russia in 1995 and pumped up its operations after the election of President Vladimir Putin in 2000 appeared to stabilize the country's chaotic economic policy. The focus of the oil company's activities is Sakhalin Island, a former penal colony off the east coast of Russia and north of Japan, which disputes Russia's sovereignty over it. Sakhalin has oil reserves of 14 billion barrels and the largest natural-gas deposits in Russia. Exxon Mobil formed a consortium with Japanese, Indian, and Russian companies to extract the fuel, retaining a 30% interest for itself. After negotiations with the Russian government that generated 41,000 pieces of paper, development began in 2001.

The project, one of the largest foreign investments in Russia, has also turned out to be one of the most complex in Exxon Mobil's history. The consortium committed $4.5 billion before pumping the first barrel last year and figures a total capital cost of $12 billion--plus. Exxon Mobil expects the venture to produce 250,000 barrels a day, which would equal about 10% of the company's global production. If Sakhalin goes wrong, the oil giant's replacement strategy would take a crippling hit.

The sheer scale of the bet is noteworthy; that it is happening in one of the world's most hostile climates makes it compelling. Conditions on the edge of Siberia are a huge challenge, with frigid, unpredictable weather and poor infrastructure and support services. There are some man-made hazards too. During the Soviet era, Sakhalin was a closed military zone, and the island is still studded with unexploded shells.

To get at the oil, Exxon Mobil has deployed the largest and most powerful onshore drilling rig in the world, an enclosed structure 23 stories tall. This is paired with an offshore platform, which has been used in Alaska and is specifically designed for use in arctic climates. Combined, the two facilities will be able to drill up to 36 wells, some of which will extend horizontally more than six miles under the seabed--among the longest such wells in the world. Getting the oil to refineries required additional layers of infrastructure and investment, including the construction of an onshore processing facility. A half-mile-long bridge connects the platform to the shore; then the crude goes into a 135-mile pipeline on the Tatar Strait, where specially designed combination tanker-icebreakers bring the goods to market.

Russian politics have blocked further efforts by Exxon Mobil to expand--and in fact are making business increasingly uncertain. The company signed a production-sharing agreement for Sakhalin-3 in 1996, but a government commission annulled the deal eight years later. Exxon Mobil was also on the verge of buying a $25 billion stake in Yukos when that company ran into political headwinds and was dismembered. Exxon Mobil plans no new investments in Russia. Even for a company that can lay down a multibillion-dollar stake in Siberia without blinking, the odds right now are just too long.

Brainpower

Except for a supersized Adirondack-style chalet that's used as a guesthouse, General Electric's Global Research Center looks like the campus of a large but undistinguished 1950s trade school. Belying its retro architectural feel, the Global Research Center, which overlooks the Mohawk River near Schenectady, N.Y., is the cutting edge of GE's effort to stay on the front lines of technological and financial excellence--and to speed up the journey of ideas from lab to market. "Create or die" is not only a cliché in the tech world; it is a natural law. (In the past couple of years, GE has also set up satellite research offices in Munich, Bangalore, and Shanghai.)

The center is growing as fast as everything else at GE these days. This year GE will spend $458 million on research into products that don't yet exist, about one-sixth of its total $3.1 billion R&D budget. A little less than half of that money is allocated by the company to long-term, big-think projects like organic light-emitting diodes (OLEDs), GE's latest attempt to reinvent the light bulb. The effort seems particularly appropriate since one of GE's founders was Thomas Edison. The Wizard of Menlo Park might be pleased that his design is still the basic template, but GE is looking for a better way. It is working on simple electrical circuits that it can stamp out on a piece of plastic. These OLEDs, GE thinks, could eventually be a more efficient alternative to fluorescent bulbs. The research to transform OLEDs from a bright idea into a flick of the switch is expected to take five to seven years--about average for the advanced technology projects at GE. The rest of the money goes to projects with quick turnaround times. For example, researchers are trying to shorten the time required to produce a high-resolution magnetic-resonance image of a patient to five seconds from 45. Another group is working on high-strength composites that can replace components in aircraft engines.

Projects range in length from a couple of months to 20 years. The trick is to keep everybody focused on goals and deadlines. Scott Donnelly, 43, the research head, who is a computer and electrical engineer by training, puts it this way: "We don't do stuff where we can't be sure where we are going. Our guys aren't confused." Among the bets that GE is pursuing, three could bring exceptional payoffs:

■ Nanotechnology: Looking into the different properties that materials exhibit at tiny scales--1/80,000 the width of a human hair--has become so sexy that six of GE's 11 businesses want a stake in the outcome. Manipulating materials at nano scales can, say, improve the durability of a tennis ball by making its cover less permeable. GE doesn't make tennis balls, so it's looking into more cosmic applications. It has put 50 scientists to work learning how to develop supermaterials, like ceramics that have greater damage tolerance, and to manipulate materials in solar cells to make them more efficient at producing electricity.

■ Biotechnology: The diseases of aging baby-boomers--Alzheimer's, cardiovascular disease, cancer--present a huge opportunity. GE is trying to figure out a way in which doctors can use molecular imaging to detect diseased cells long before there are any physical signs like tumors or growths. GE has built new lab space as part of a $100 million renovation at the center and plans to add another 30 to 40 researchers to the 50 it already has working.

■ Sustainable energy: Sanjay Correa, GE's global technology leader for energy, calls the hunt for clean, sustainable energy "the biggest question for mankind in this century." GE has 300 researchers investigating a gaggle of technologies, from hydrogen and photovoltaics to clean coal and wind. One-sixth of that scientific army is focused exclusively on hydrogen: how to make it, store it, and transfer it so that it can be converted into electricity. "We know things will change," Correa says. "The future is happening faster than anyone thought."

And GE intends to grab a big chunk of it.

FEEDBACK ataylor@fortunemail.com

REPORTER ASSOCIATES Dana Castillo, Jia Lynn Yang