(FORTUNE Magazine) – Twenty-five floors above San Francisco Bay, on the teeming trading floor of an investment bank, John Walecka eagerly awaited what venture capitalists live for--the big IPO. In this instance, a high-tech startup called Xylan was going public. For Walecka it would be the payoff from a bet on a team and a dream. Over the past two years the lanky, 36-year-old former Stanford oarsman had pulled hard for Xylan, serving as key recruiter and chief money raiser. When Walecka came aboard, the entire staff could squeeze into a racing shell. The entrepreneurs had barely unpacked when the L.A. earthquake ravaged their offices. But the crew didn't miss a stroke, even cloistering themselves in a cheap hotel to conceive their brainchild, a powerful new switch for networks of PCs and workstations.

Before long, the switch had become a sensation, and on this March day Walecka expected a sumptuous reward for his investment of passion, sweat--and $4 million. He was hoping Xylan would open somewhere above its official offering price of $26. As Walecka scanned a bank of screens for the first quotes, the traders at Robertson Stephens squeezed by him in the narrow aisles, buzzing about a big debut. Then the Nasdaq symbol XylanCp flashed for the first time. Walecka gaped in tousle-haired glee at the second seismic event in Xylan's brief existence: The ticker read $55 a share, giving Xylan, a company with sales of $30 million, a market value of $2.3 billion. Raved Walecka: "This is the deal of a lifetime!"

Try all-time. Move over, Netscape: Based on the first day of trading, Xylan stands as the most profitable venture deal ever. (Today it's selling at around $64.) Walecka's firm, Brentwood Associates of Los Angeles, reaped $390 million in profits, a 5,000%-a-year return on its puny investment. Its winnings dwarfed the $250 million rival Kleiner Perkins earned on Netscape's opening day last August. Though both companies debuted at around $2 billion, Brentwood owned a far bigger chunk of Xylan--17%--than Kleiner held in Netscape.

That was music to Brentwood's investors, chiefly pension fund managers like those at Brinson Partners, which put just $80 million in the pool that backed Xylan and 27 other startups, like software maker Documentum, another hot 1996 IPO. Xylan alone returned the entire fund five times, the kind of return you can usually get only from a winning lottery ticket. Quipped a dazed Walecka as he departed Robertson Stephens: "If I don't get back to work, my partners will ask, 'What have you done for me lately?' "

What the IPO market has done for venture capitalists these days is nothing short of astounding. Walecka and an elite group of others in Silicon Valley like John Doerr of Kleiner Perkins, who brought Netscape public, and Michael Moritz of Sequoia, the power behind the latest hot IPO, Yahoo, are profiting wildly. In the 1980s a major venture capital success returned perhaps 20 times the investment in, say, five to six years. Now the homers harvest 30 to 100 times the outlay in as little as three years. To be sure, each Netscape supports a flurry of bombs. But the miss rate is surprisingly low. Traditionally, about three investments in ten go sour, one to two score home runs, and five or six bump along, finally finding buyers at a modest profit or loss--in today's market, more likely a profit.

The result: returns seldom seen since the LBO funds of the late 1980s. For instance, Ann Winblad's firm, Hummer Winblad Venture Partners, which backs software startups, has been boasting almost 50% annual returns to investors over the past six years. Some newer venture funds are doing even better. Started in mid-1993, the $135 million pool managed by Accel Partners--its IPOs include UUNet, a service firm that links companies and individuals to the Internet--is cooking at over 100% a year.

Who exactly is partaking in these huge returns? Venturers get their money chiefly from pension funds, college endowments, and insurance companies. Those once cautious institutions are developing a yen for the game. Last year they poured $4.5 billion into venture funds, more than twice the record set in 1987. Today aggressive pension managers at Nynex, AT&T, and the state of Michigan can park as much as 4.5% of their assets with venture capitalists, twice the typical share in the 1980s; the Stanfords and Princetons place from 10% to 15% of their endowments in LBO and venture funds. Overall, America's venture pools--which invest in everything from medical supplies to restaurants--far outpaced the Nasdaq and S&P 500 by returning 21% a year. And the top Silicon Valley firms paid far more.

The profits are flowing because these venture capitalists are both smart and lucky enough to be riding an unprecedented economic wave. First, the revolution that's allowing America's 72 million PCs to talk to each other is opening a host of lucrative new fields, from network switching to client-server software. At the same time, those institutional investors that have been filling the venture capitalists' coffers have made it possible for them to back an impressive number of startups in those new industries. Last year venture firms invested $7.5 billion in young companies, 50% more than in 1994, according to VentureOne, a San Francisco research organization. Many venturers think the banquet will last forever. "The world has changed," says Doerr. "We've never seen so many great entrepreneurs and technologies." As Paul Wythes, a partner in Sutter Hill Ventures, says: "This is the golden age."

Just who are these ultimate capitalist gamblers? The community is surprisingly small, even incestuous. Most of them work on or near Sand Hill Road in Menlo Park, California, a medley of low-slung office buildings nestled in a rolling countryside that forms a swarming anthill of money. The venture capitalists frequently invest in each other's deals and mingle at events like the apres-football tailgate parties at Stanford, a favorite alma mater.

A bunch of them cherish nothing more than rolling shoulder to shoulder at the Las Vegas craps tables. Twice a year Walecka gathers with other hot young venturers like Geoff Yang of Institutional Venture Partners at Las Vegas computer shows for a male bonding ritual: rollicking sessions around the poker, craps, and blackjack tables at Bally's. The crew is so familiar that croupiers call Yang "Mr. Vegas." When the bleary-eyed brigade emerges from Bally's, the sun is rising over the desert. "It's a lot like my profession," says Walecka. "Only, our odds are better."

As the Vegas folk might say, these guys are on a rush. The group profiled here range in age from 35 to 45. With the exception of Michael Moritz, they're all technology jocks, usually with hands-on training at an Apple or Intel. They adore nothing more than chewing over the latest wrinkle in, say, the asynchronous transfer mode. Their strongest bond, however, isn't science but faith, a yen for gambling on a concept before seeing a product or even a business plan. It's an ability to separate the Rube Goldbergs from the Steve Jobses of the world.

MR. ROLODEX. To blossom, a great idea needs a strong team. No one is better at assembling a superb crew at shimmering speed than John Doerr. Over the years, he's brought Sun Microsystems, Intuit, and Netscape public--all high-tech mega-smashes. Doerr's a combination science geek and whirlwind salesman. Visually, the former image predominates. Skinny and fidgety, his narrow face bridged by thick wire-rims, Doerr talks in a staccato basso and compulsively downs handfuls of popcorn. He drives as erratically as a dog looking for a place to relieve itself. An engineer who once sold Intel chips and software to Indiana farmers, Doerr swings down a hallway, his pager beeping and mobile phone ringing--a walking communications center. He unwinds doing predawn laps in the chilly water of his outdoor pool.

His zealotry extends to ideas. "As a salesman, he's so good he can sell you just as easily on bad concepts," says Netscape Chairman James Clark. Doerr has shot a number of arrows into space, including a company called Go, a disastrous foray into pen-operated computers. His weakness is betting on futuristic products with little evidence consumers want them. Customers, for example, mostly yawned at pen computers. "John thinks you can manufacture markets wholesale, and you can't," says Ruthann Quindlen, a partner at IVP.

His big successes came in companies like Intuit, Sun, Compaq, and Netscape, where a budding market craved good management and innovative products. At Netscape, Doerr bought a solid vision devised by somebody else, a new browser to guide the waxing throng of surfers through the World Wide Web. Founder Clark showed Doerr the idea in early 1994 when two other firms, Mayfield and NEA, were balking at the price of the deal. Doerr didn't quibble over the price of $5 million for roughly 13% of the company. "The Internet will be three times bigger than the PC," raves Doerr.

The challenge was speed, launching a high-quality browser before the competition could dominate the market. A core engineering team, led by the then 22-year-old programmer Marc Andreessen, was already in place. But the startup had no real management, and Clark didn't want to run the company forever. Doerr's mission: help recruit a full team, and a world-class CEO, in less than 120 days.

The Netscape atmosphere--a dormitory-like conclave of University of Illinois programmers who slept on futons and toted boom boxes--was catnip for Doerr. Tapping his vast Rolodex, as well as his Kleiner partners, Doerr found marketing chief Michael Homer, a veteran of Apple as well as Doerr's pen-computing fiasco Go. Doerr, working with partner Kevin Compton, found sales manager Todd Rulon-Miller, then ceo of a California software company. "At the interview, John made me recite, name by name, who I'd hire to get into 5,000 retail stores by January 1," recalls Rulon-Miller. "He also wanted to know who would head OEM and direct marketing, and how fast I could hire them." Eight days after Doerr's first call, Rulon-Miller started at Netscape.

The Doerr mystique proved a powerful recruiting tool. Doerr accompanied Clark to the first meeting with James Barksdale, who had been president of McCaw Cellular and is now Netscape's CEO, over lunch at a Red Lion hotel in Seattle. "He didn't know me from Adam," recalls Clark. "It was Doerr who intrigued him and played a big part in getting him to sign up."

THE STARMAKER. Few venture capitalists are better at finding promising entrepreneurs than Michael Moritz, whose firm, Sequoia, backed such legendary successes as Apple and Cisco Systems. He is a rarity in the venture game, a blithe spirit who relishes books on Winston Churchill as much as network switching. Son of a classics professor from the University of Wales, Moritz, 41, edited the famous literary magazine Isis at Oxford, then pursued an MBA at Wharton in 1976. Moritz enjoyed lounging in Philip Roth's creative writing class far more than discounting cash flows. On graduation, he took a job in Detroit, not as assistant treasurer at Ford, but, gulp, as a correspondent for Time. "It sounded like more fun than selling bonds or pontificating for McKinsey," says Moritz. "A week later I was on the Chrysler jet with Lee Iacocca."

After covering Silicon Valley for Time and writing a stinging, unauthorized book on Apple Computer, Moritz joined Sequoia in 1986, guiding startups like modem and software company Global Village, a big IPO in 1994. How did he learn so fast? First, the swarthy, black-haired Moritz follows a simple philosophy--serve an existing market, don't try to invent one. Says he: "This game is not big-bang creationism."

Beyond that, Moritz has a distinct profile for the founders he backs. He prefers young, hungry first-timers: "Entrepreneurs peak early, like mathematicians and swimmers. The young ones are too fearless to worry about the future." He also favors fighters who surmounted tough backgrounds, especially immigrants from India, Korea, or Israel. Quips Moritz: "The names in our stable sound like the delegates to the U.N."

Moritz also prizes founders who are tight with a dollar. His inspiration is partner Don Valentine, 63, a penny-pinching legend who likes to tell the story of the Cisco executive who once plucked a nickel from a urinal. Sequoia's offices wear AstroTurf-thin carpeting and cheap furniture like a badge of honor. Moritz gives his founders the minimum for survival. "We keep them on a short leash," he says dourly. "With too much money, they go soft."

At Yahoo, Moritz discovered the kind of entrepreneurial soul he's always looking for. Early last year an Internet-surfing friend told him about a hip new Web guide called Yahoo. Moritz found that the service wasn't a business at all but a hobby for two engineering grad students, David Filo, 30, and Jerry Yang, 27. Moritz made a pilgrimage to their office, a trailer located behind the Bill Gates building on Stanford's campus. To avoid drawing attention, Yang and Filo told all their visitors to come sans ties so they'd look less like moneymen. Moritz found the inventors surrounded by smelly running shoes and stray golf clubs, windows thrown open to cool the radiator-hot computers. "It was like the Black Hole of Calcutta," sniffs Moritz.

For Moritz, Yahoo not only tapped the Internet craze, but was a cheap bet. With no marketing, the guide already drew a throng of loyal fans. Besides, the founders fit the Moritz mold. Born in Taiwan, Yang grew up with his mother in a scruffy neighborhood of San Jose. "I had nothing to lose," says Yang. He and Filo are as tightfisted as they are fearless. To this day, Filo drives a rusty 1981 Datsun and calls himself "the cheap Yahoo." Moritz loves it. "The founders set a frugal tone," he muses. "Corporate America wouldn't know what to do with these beasts."

As usual, Moritz offered the minimum: $1 million for 25% of Yahoo. But the founders had job offers from both Netscape and America Online padded with some stock. Moritz appealed to their sense of adventure. "He said, 'If you join a big company, you'll never know what you could have done on your own,' " recalls Yang. He also convinced Yang and Filo that by running lean, they could keep a huge chunk of the equity, and profit handsomely if they ever went public. The founders chose freedom, as Moritz puts it, over "indentured servitude." "He never pretended his end game wasn't financial," says Yang. "But he also showed us that building our own company could be a growth experience."

Moritz's advice was prescient. Yahoo, supported by ads for everything from cosmetics to software, owns one of the marquee brands in cyberspace, serving one million users a day. On the day it went public in April, its shares shot up from $13 to $33, which adds up to a market value of $850 million--not bad for a company formed a year ago. As for Yang and Filo, they're each worth $132 million, a shade below Moritz's Sequoia stake of $146 million.

CAPITALIST OF THE WILD WEST. Amazingly, even the most creative founders can target the wrong market. Often, it's the venturers who help them re-aim--by totally changing products. That's just what Geoff Yang of IVP did with startup MMC Networks. Back in 1994, the Santa Clara, California, company had been struggling with its concept--a new entry in the ranks of network switches that let PCs and workstations trade image, voice, and data, as well as handle heavy Internet traffic, at ever higher speeds. By prodding MMC to focus not on the switch itself but on the processor at its core, Yang vastly enhanced the fledgling firm's prospects.

That job took all of Yang's skills as an engineer, a businessman, and especially a diplomat. Son of immigrants from China, Yang majored in information systems at Princeton and took the full course load for an economics degree as well, while working evenings at a savings and loan bank. His 1981 thesis, bridging both departments, led to a successful new product for his employer, a negative-amortization mortgage for young homebuyers. Meanwhile, Yang, a net-rushing tennis star, ran a restringing business: He charged $18 a racket, paying students $5 to do the work.

After college, Yang discovered that corporate life wasn't for him. During two years as an IBM salesman in New York, he liked the gung-ho spirit but deplored the bureaucracy. When training class opened with a chorus of "Take Me Out to the Sales Game," no one sang louder than the irrepressible Yang. "But it wasn't a meritocracy," he grouses. "IBM would force the smartest guy on the planet to plod ahead slowly, two years in each job." After finishing his MBA at Stanford in 1988, Yang chose adventure. "Venture capital is the Wild West!" he exults.

His Chinese superstitions, claims Yang, register when he's nearing a good deal. When first visiting MMC in 1994, he rejoiced to find its offices a floor below a company he had helped sell at a large profit. "The spirits were right!" says Yang. But he found the strategy all wrong. MMC's chief engineer, an Israeli immigrant named Alex Joffe, had designed a new switch processor, the set of chips that efficiently organizes the data that flow through a computer network. The processor solved a thorny problem by increasing the network's speed without incurring the high costs of adding extra memory.

But founder Amos Wilnai, a graduate of Israel's MIT, the Technion, wanted to produce not only a processor but an entire switch. Yang argued that MMC innovations were mainly in the chip architecture. Making switches, he argued, would require a bigger up-front investment, both to produce and market. And he didn't relish competing against giants like Cisco. "I told them, 'Cisco will kick your butt,'" recalls Yang.

By specializing in processors, Yang argued, MMC would have the Ciscos of the world as customers rather than competitors. Also, it would be the first player in an entirely new industry. "The leaders always get the biggest market shares," says Yang. "Those shares get set early, and don't change much."

But Yang's solution required a leap of faith. Most switchmakers like Cisco claimed they wanted to keep designing their own processors.Yang felt strongly, though, that in the future, more and more companies would outsource everything but the few features that truly set their products apart. Besides, Yang couldn't see why the switchmakers would want to keep designing their own processors, since MMC's was poised to become an industry standard anyway.

Yang prevailed. "Over time," says he, "the Ciscos realized that, for them, silicon is sand." Adds MMC's Joffe: "Yang was right. Our core experience was in chip design." Today Cisco, CrossCom, and two dozen other manufacturers are building the MMC chip into their switches. Brimming with orders, the company is eyeing an IPO in 1997.

THE CIRCUS ACT. As Doerr's experience with Go and other failed startups suggests, fresh technologies are futile unless hungry buyers await. That's a lesson Ann Winblad of Hummer Winblad holds close to her heart. "We invest in markets, not technology," says Winblad. "And we want opportunities the size of Texas." While most venturers invest in several high-tech areas, Winblad does only in software, learning every bluff and stream in her territory. The 5-foot-3 Winblad and her 6-foot-9 partner, John Hummer--once an NBA center--form the venture world's odd couple. "When we walk in to meet entrepreneurs," jokes Winblad, "they think the circus has just arrived."

Winblad is a child of software. A high school cheerleader and valedictorian from Farmington, Minnesota (pop. 2,100), Winblad started Open Systems, a supplier of accounting software, on $500 in 1976. Seven years later she sold it for $16 million. Around that time, she started dating the industry's prince, Microsoft's Bill Gates. The couple shared cerebral "theme" vacations unrelated to the places they visited. For Brazil, it was biotech. Winblad and Gates sat side by side in a sailboat, reading his-and-hers copies of a two-volume work on the molecular biology of the gene. In Santa Barbara they watched a 20-part video series on physics.

Winblad's marketing skills nurtured one of 1995's hottest IPOs, Arbor Software. The company's programs enable businesses to prepare sales forecasts and budgets using numbers buried in PCs sitting in far-flung offices. In 1990, when founders James Dorrian and Robert Earle first broached the idea of crunching business data on PCs, most venture capitalists balked. At that time, companies relied mainly on mainframes. Worse, Dorrian and Earle, then computer consultants to the city of San Francisco, weren't exactly well-known programmers. Incredibly, Dorrian learned to write his own code, absorbing the basics cookbook-style through manuals.

Winblad doubted the neophytes could deliver. But unlike most of her rivals, she saw a big market if they did. She took Dorrian and Earle to visit potential customers, mainly some information technology managers she knew at big companies, including Sun Microsystems. In each place the inventors made their pitch, then left the room while Winblad debriefed the audience. "At Sun," recalls Winblad, "the response was, 'If this was real, we'd have bought it yesterday.' "

Winblad then told Dorrian and Earle to come back if they could build a prototype. The pair hit bottom: Dorrian sold his house, and Earle declared bankruptcy. But in 1991 they showed Winblad a rudimentary program. A week later she and another early believer, Sequoia, invested about $1 million each.

Always the adroit marketer, Winblad demanded that they change the company's name from Groupware International to Arbor. "One mistake," she says, "and they'd be forever known as 'Gropeware.' " During the roll-out in 1992, Winblad clashed with Dorrian and Earle. The founders wanted to race ahead by hiring a big, costly sales force, a move that would require more capital. Winblad preferred building orders gradually, by reputation, the way movie producers create excitement by releasing a film in a few select theaters and then expanding onto more screens only when fans who've heard about it scream to see it.

Winblad's strategy won. Arbor courted key accounts like Sears, which credited Arbor with cutting its data-processing costs by tens of millions of dollars. The Sears-style endorsements won a slew of new business. Today Hummer Winblad's stake has mushroomed to $100 million. The founders--who once missed car payments--are worth $50 million apiece.

Can the ideas and money keep flowing forever? That's the question that's beginning to haunt some venture capitalists, even as they are enjoying one of the hottest streaks in the history of their industry. For in all likelihood the good times will end, and probably sooner than many think.

What's made each of these venturers so wealthy is that almost every one was an early player in one of the new, high-tech markets--Internet navigators, for instance--that have been sprouting up. In these markets, leaders like Netscape grab a large share of business and enjoy fat margins. That makes for highly attractive IPOs.

Now, lured by the tremendous returns that going public can bring, new challengers are gearing up to crowd these fields. The late-comers, featuring me-too products and weaker brands, will be far less profitable than the companies that were there first. Hence, the IPOs to come aren't likely to generate nearly as handsome returns. Nor will life be as charmed even for entrepreneurs who are pioneers in new markets. As more and more companies learn to design and build products at an accelerating rate, the early innovators will own the market for shorter and shorter periods. That, too, will knock down venture returns.

In the meantime the gold rush keeps going, and the venturers, the rawest gamblers in all of capitalism, continue to put fortunes on the line. Says Geoff Yang: "I've never seen anything like it. Things are so good I can't imagine them getting any better." Maybe he's right. If so, the question then becomes: When will it get worse?