(FORTUNE Magazine) – Back when all you needed was trust and a little pixie dust, scores of startups had grandiose plans for making vast fortunes. Now, of course, many of those companies--and the people who invested in them--have lost their illusions along with their money. But if there's one firm that epitomizes the "I believe" mindset so prevalent on Wall Street in the late 1990s, it's Winstar, the former ski apparel shop that convinced the world that it could fly as a broadband company.

The sheer size and head-snapping speed of Winstar's wipeout help set it apart from other high-tech misadventures. In March 2000 its common stock hit a high of $65, giving the company a market value of about $10 billion. A little more than a year later, on April 18, 2001, Winstar filed for bankruptcy. Trading in its common stock was halted at 14 cents a share, and much of what Moody's reckons to be $6.3 billion in debt is now probably worthless.

Winstar also stands out because its believers weren't just day traders armed with E*Trade accounts. They were blue-chip investors, the kind that in an earlier, more innocent age were known as "smart money." Microsoft was a major investor, as was top private-equity firm Welsh Carson. The private-equity arm of Credit Suisse First Boston sank almost $400 million into Winstar. Equipment sellers including Cisco, Compaq, and especially Lucent, which as of early April was owed around $700 million by Winstar, joined the party too. Smart money, evidently, has shed a few IQ points over the years.

Now Winstar is depicted as a casualty of capricious capital markets. "Last year the market was willing to invest $1 today for $2 tomorrow," says David Barden, who followed Winstar for J.P. Morgan. "Today the market won't lend you a quarter to buy a diamond mine." In no way can Winstar itself be held responsible--or at least that's the view of the company, which in fact blames its predicament on Lucent, which it is suing for $10 billion for alleged failure to deliver on "promises of expertise and financial support."

But it's ludicrous to think of Winstar solely as a victim in all this. It was an exceedingly eager participant in a system that rewarded companies for keeping up appearances at any cost. And analysts and investors were too blinded by greed, too captivated by an alternate financial reality known as Ebitda, and too tangled up in conflicted interests to raise questions--much less look at the company's balance sheet. Both Winstar executives and many of the Wall Street analysts who followed the company--including CSFB's Mark Kastan, who maintained his $79 price target on Winstar until April 6--continued to hype the stock as it slid to less than $1. "Analysts are always pitching for the next deal, so they want to pull the wool over their own eyes," says a former Winstar analyst.

But bashing the analysts for their failure to analyze is too easy. After all, they were just saying what almost everyone, from chat room investors to mutual fund managers at firms like Janus (which owned over 10% of the company's common shares at year-end) to Winstar's backers, wanted to hear. And even the supposedly sophisticated people chose to ignore the wildly flapping financial red flags telling them that Winstar, particularly in a grim market, was not going to fly.

Now that we're back from never-never land (or so we hope), it seems surprising that the stock market ever saw Winstar and its peers--so-called competitive local exchange carriers, or CLECs, which sprouted in the wake of the Telecommunications Act of 1996--as a sure bet. The idea behind most CLECs was to spend billions building last-mile networks to compete with the Baby Bells. It was an expensive proposition, but if a CLEC could capture enough business, AT&T or WorldCom just might pay a huge takeover price to rid itself of the competitive nuisance. CLECs weren't supposed to produce profits until some distant date. Even bullish analyst Kastan didn't expect Winstar, which provided broadband services to office buildings using a high-frequency wireless spectrum, to make money until 2007.

Of course, there was no proof that Winstar would ever have enough customers to support the billions in investment. You had to go on faith--which was easy for the Wall Street analysts, whose companies stood to profit handsomely from Winstar's immense investment-banking needs. From 1995 through 2000, Winstar raised billions in capital. The fact that the company's CEO, Bill Rouhana, a former financier for telecom and media companies, had a huge stash of pixie dust--and knew just where to sprinkle it--certainly didn't hurt. "There was always a lot of flash and pizzazz at Winstar," recalls Glenn Waldorf, an analyst at UBS Warburg.

But it wasn't just the analysts who had much to gain from Winstar. In October 1998, Lucent agreed to lend up to $500 million at a time to fund the buildout of what Winstar grandly described as the "world's first global end-to-end broadband network." That so-called vendor finance arrangement allowed Lucent to bolster its own revenues by selling equipment to Winstar (the specifics of the deal, which we'll get to a little later, were highly unusual, to say the least). Then, in December, Williams Communications said it would pay Winstar $400 million over the next four years to use its wireless broadband network. In return Winstar agreed to pay an even bigger amount back to Williams for the use of its fiber-optic cable--$644 million over seven years.

All the while, Winstar was losing buckets of money and stepping up its borrowing. In 1999 its net losses jumped from $488 million to $700 million, and its debt ballooned from around $2 billion to more than $3 billion. But never mind. Winstar's revenues had almost doubled, to $446 million. Besides, broadband was the next big thing, and everyone--from private-equity firms to other technology outfits--wanted in on the action. Early last year, with its stock hitting new highs almost daily, Winstar sold around $900 million worth of convertible preferred stock to Microsoft, CSFB, and Welsh Carson.

Winstar took advantage of the fact that others were eager to lend it money. Some of its existing debt required the company to meet certain operational targets--for example, a path to profitability. So in March, Winstar replaced that debt with a fresh $1.6 billion in high-yield debt carrying less onerous terms. The offering was led by Salomon Smith Barney and CSFB, which earned what one observer calls "really exorbitant" fees--more than $50 million, according to Securities Data.

Winstar by then was also bumping up against its borrowing limits with Lucent. So in May the company pledged nearly all of its current and future assets (excluding those financed by Lucent) as collateral for a $1.15 billion loan from Bank of New York, CIBC, Citicorp (Salomon Smith Barney's parent), CSFB, Credit Lyonnais, J.P. Morgan, and others. Winstar paid off Lucent, which promptly offered more money--up to $1 billion at a time in financing. (Lucent, still trying to meet Wall Street's revenue targets, wanted Winstar to keep buying its equipment.) At that point Winstar declared that it was funded through 2001, an assertion dutifully repeated by analysts like Credit Lyonnais' Rick Grubbs, who initiated coverage of the stock on June 23, when it was about $40, and set an $86 price target.

When the market started to quake last summer and through the fall, Wall Street suddenly began to worry about the old-fashioned notion of earnings. But that newfound concern didn't extend to CLECs like Winstar, which weren't close to making money. Instead the Street was focused on a more attainable measure--Ebitda, or earnings before interest, taxes, depreciation, and amortization. Often referred to as "operating cash flow," Ebitda is a metric used by all sorts of companies (including FORTUNE parent AOL Time Warner). But purists hate it because, as Winstar's own filings note, "Ebitda is not intended to represent results of operations or cash flows from operations determined in accordance with accounting principles." In other words, investors need to scrutinize Ebitda carefully to make sure it reflects the real cost of running a given business. Nor does "positive Ebitda" necessarily mean that a company can cover the interest payments on its debt or fund its capital expenditures.

Nevertheless, with Ebitda as its yardstick, Winstar was able to please the Wall Street analysts. Its losses narrowed--and just as important, they were less than analysts expected in both the second and third quarters. "Winstar is demonstrating strong execution here, which has become increasingly important in validating companies in the CLEC sector," wrote Salomon's Jack Grubman, who also noted that Winstar was "fully funded."

In fact, Grubman and the rest of his sell-side brethren seemed to ignore Winstar's spreading and potentially company-killing cancer, which was growing right there for all to see on Winstar's balance sheet. By the end of the third quarter, Winstar's accounts receivable had expanded to $230 million, a 70% increase from year-end 1999 and a sign that Winstar customers might not have been paying up. Debt, if you included preferred stock, had jumped to around $5 billion. But under current accounting rules, companies don't list all preferred stock as a liability at full value. So Winstar's balance sheet reported negative shareholders' equity of $110 million, when if you had included an additional roughly $1.3 billion in preferred stock, the number would be much higher--about negative $1.4 billion. In discussing Winstar's valuation, analysts acted as if those pesky preferred shares didn't exist, despite the fact that their owners would have to be paid in full before common stockholders could claim 1 cent.

There were other oddities tucked away in Winstar's New Age books that might have troubled analysts less engaged in cheerleading. During 2000 and early 2001, Winstar invested a total of $145 million in a now near-bankrupt B2B company called Wam!Net--including $95 million in cash that you might think it would have wanted to conserve. There was method behind that seeming madness. Wam!Net, a Winstar customer, paid the company an initial $20 million and promised to make quarterly payments beginning at $5 million and increasing to $25 million over the next seven years for its use of Winstar's broadband capacity. Without Winstar's investment Wam!Net would have been hard-pressed to keep paying, and Winstar's all-important Ebitda would have suffered.

Wam!Net wasn't the only questionable transaction boosting Winstar's Ebitda. On the last few days of the second and third quarters--what timing!--an Internet services company named Savvis (which, by the way, is also on the verge of bankruptcy) agreed to purchase $30 million and $10 million, respectively, of equipment and services from Winstar. Curious about how Savvis and Winstar might have found each other? Both were financed by Welsh Carson. (Winstar says the two were introduced by Savvis' bankers.)

Supposedly Winstar's core business was hooking up customers to its broadband network, not making one-time sales of capacity and equipment to other providers. But, like other CLECs, the company didn't break out its revenue--and if the analysts bothered to ask, they certainly didn't broadcast the answer. In addition, deals such as Winstar's arrangement with Wam!Net, Williams, and Savvis allowed the company to record revenues. But at least some of the costs of those revenues--for instance, Winstar's investment in Wam!Net or its payments to Williams--aren't accounted for in Ebitda but are booked on the balance sheet. So arguably, the true operating cost of the business is understated. Skeptics also suspect that Winstar and other CLECs were able to use vendor finance arrangements (like Winstar's contract with Lucent) to record operating expenses on their balance sheets--expenses that would otherwise reduce Ebitda. "The only thing good about Ebitda is that the Street wanted to see it, and if Winstar gave it to them, they could raise more money," says one analyst.

More money, of course, is exactly what Winstar badly needed. Sure, shrinking Ebitda losses implied that the company was losing less money than expected, but the cash the company had raised also disappeared much faster than most anticipated. In November, Microsoft, CSFB, Welsh Carson, and a new backer--Compaq--coughed up another $270 million for more preferred stock. In addition both Cisco and Compaq agreed to lend Winstar money to purchase equipment from them. Siemens Financial, a subsidiary of the huge German electronics firm, lent Winstar another $200 million. (On Dec. 15 a press release announced that Winstar would purchase $150 million of equipment and services from Siemens Carrier Networks over the next two years.) "Company funded into 2002," crowed Winstar's press release and analyst reports. Deja vu, anyone?

That's not to say there weren't any Winstar doubters. By the end of 1999, 25% of its shares were sold short. In early December the doubts broke into public view when Goldman Sachs' high-yield analyst Mark Rose downgraded Winstar's bonds to market performer--a Wall Street euphemism for the unspeakable "sell." In a written report, he noted that Winstar ranked at the bottom of its peer group, "owing to its highly leveraged capital structure." Rose also noted that Winstar's net debt was 4.2 times its annualized revenues, an alarmingly high ratio and more than its total investment in property, plant, and equipment.

In early January, CIBC analyst Cannon Carr broke from the pack and became one of the few equity analysts to put a hold (yes, that would also be sell) rating on the stock. "Winstar is a client of ours too," he says now. "But look: The facts are the facts." Among other things, he noted that the company would owe some $400 million in cash interest in 2001. "Everyone was focused on income statement metrics," says Carr. "But the more you looked at the balance sheet, the less healthy it looked." Carr's criticism notwithstanding, by January's end Winstar's stock had almost doubled to $19, helped by happy talk from Salomon Smith Barney.

Soon after, though, rumors began swirling that Lucent, by then desperate to do whatever it took to polish its own balance sheet, was reluctant to turn over more funds to Winstar. By the time Winstar announced year 2000 results in late February, its stock had sunk back to $11. Even so, the bulls stubbornly celebrated Winstar's numbers. Ebitda losses had fallen from $297.3 million in 1999 to $153.4 million. (Never mind that bottom-line losses ballooned to more than $1 billion.) Again, Winstar stressed that it was fully funded through 2001 and denied that its relationship with Lucent was in trouble. "Better than expectations," raved Salomon's Grubman, who called the stock "severely undervalued" and reiterated his $50 price target.

Anyone who was silly enough to follow Grubman's advice didn't get to stay in never-never land for long. In early March outspoken short-seller Manuel Asensio released the first in a series of devastating reports saying that Winstar's common stockholders had a "far greater probability of losing their entire investment than they realize." Asensio was the first to assert publicly that the company's Ebitda had been pumped up by putting operating expenses on its balance sheet and selling equipment and services to related parties. "Winstar is like a Broadway sketch," says Asensio, who dismisses the analysts as "idiots."

Asensio also pointed out another previously ignored snippet of information that could be bad news for common stockholders. The two issues of convertible preferred stock bought by Microsoft, Welsh Carson, and CSFB contained an odd provision, one that some call a "death spiral" because it works like this: As shares of common stock decline in value, preferred shareholders have the right to convert their shares into an ever-increasing percentage of the company's equity, unless the company can pay them back in cash. According to one analysis, with Winstar's price at $10.25, preferred shareholders could eventually squeeze the common stockholders' ownership of the company down to less than 50%.

The response from the analyst camp? Buy! Salomon and CSFB, Winstar's bankers on its high-yield deal, were the company's most vociferous defenders. "There are not serious default concerns on [Winstar's] bonds," said Grubman on March 8. He characterized Asensio's report as "incomplete, inaccurate, or inconsistent with our analysis." For his part, CSFB's Kastan emerged from a March 13 meeting with Winstar management parroting the party line. He called the get-together "upbeat," said the concerns were "overdone," and quoted company executives as saying that Winstar had "significant headroom" on all the covenants surrounding its Lucent debt. Kastan repeated his $79 price target on the $7 stock.

Skeptics who bothered to run the numbers couldn't understand why the analysts were so confident about Winstar's funding status. By the end of 2000, Winstar's accounts payable had leaped to $457 million, or over 370 days outstanding in the fourth quarter, vs. a normal accounts payable period of no more than 90 days. Surely antsy vendors were going to want their money? Once Winstar had paid off vendors and covered its payroll, it wasn't at all clear that there would be enough left over from the $316 million cash the company had at year-end to pay interest on its debt--and Winstar had a $75 million interest payment due on its high-yield notes on April 16. Asensio argued that even if Winstar produced the cash, the banks that had lent it $1.15 billion last spring weren't going to let Winstar pay money to other lenders if they had any doubts about getting fully paid themselves.

Clearly the banks were getting nervous. Although Winstar's secured bank debt doesn't trade publicly, in mid-March Asensio reported it was changing hands at around 70 cents on the dollar. That indicates that the banks had calculated Winstar's total value at under $1 billion, far less than the roughly $7 billion implied by the company's common stock, which was then going for $6--let alone the some $13 billion that Kastan was still telling investors Winstar was worth.

On Monday, April 2, Winstar announced that its 10-K would be delayed--never a good sign. Merrill Lynch analyst Ken Hoexter finally downgraded the stock, which fell more than 50% that day, to 88 cents. (Thanks, guy!) The next day Glenn Waldorf, a new analyst at UBS Warburg, made his pessimism clear by rating Winstar a hold. "The reason to value equity is cash flow," he says. "And how is this company ever going to generate enough money to pay the interest on its debt, let alone the principal?" A few days later Winstar said it would lay off almost half of its 4,500 employees and curtail expansion plans. Kastan simply dropped coverage of the stock. Late on April 17, with shares at 14 cents, Winstar announced that it couldn't make its $75 million interest payment. Grubman helpfully downgraded the stock to an underperform, noting that "at this point it makes no sense to continue our optimistic stance."

The next day Winstar announced that it was filing for bankruptcy and immediately fingered Lucent as the responsible party. In its lawsuit, announced the same day, Winstar contends that the filing is the direct result of Lucent's refusal to hand over $92 million--money Winstar says it was owed under the terms of their agreement. And Winstar needed that cash to make its interest payment. Lucent, which calls the $10 billion suit "frivolous," contends that Winstar is the one that defaulted on their agreement. In its most recent quarter Lucent took a 15-cents-a-share charge to write off its loans to Winstar and other losses.

It's anyone's guess how the courts will decide. But both Winstar and Lucent are guilty of mutual exploitation. Winstar's contract with Lucent was extraordinary, even by the standards of such "vendor finance" deals. In most such arrangements the company would lend maybe 10% or 15% above the cost of the equipment it was selling. Not only did Lucent hand over quite a bit more--more than 40% above the cost of the equipment--but it also financed Winstar's purchases from other companies. In other words, Lucent was acting more like an aggressive bank than an equipment seller.

How will this tale end? In a press release, Winstar is sticking to its relentlessly upbeat script. It claims that "we expect to emerge from the Chapter 11 process with a new balance sheet"--in other words, this is simply the beginning of a new, improved life, business as usual except without all that nasty debt. Then again, Winstar's lawsuit notes that unless Lucent is forced to pony up, Winstar will "face the possible destruction of its business." At least three lawsuits alleging that Winstar made "material misrepresentations and omissions of material facts" have also been filed against the company. But none of the Wall Street analysts have reiterated their "buy" ratings--yet.

FEEDBACK: bmclean@fortunemail.com