CHAPTER 4

Ravenswood Winery, Sonoma, 1999

Bill Gurley moved to Silicon Valley in 1996. The six-foot-nine Texan was a Wall Street research analyst who had just been hired by Frank Quattrone, a big-name investment banker and the former leader of Morgan Stanley’s technology practice. Quattrone had left to build something similar at Deutsche Morgan Grenfell (DMG).

When Quattrone’s offer came, Gurley was working at Credit Suisse First Boston. But the analyst had his sights set on becoming a venture capitalist. He made Quattrone promise him a transfer to Silicon Valley. Quattrone not only agreed but told Gurley he would introduce him to all the venture capitalists he knew if he joined Quattrone’s team.

Under the model that Goldman Sachs had designed the previous decade, investment bankers and analysts worked hand in hand to market IPOs. While the bankers worked on crafting the company’s story and writing the prospectus, the analysts learned the business model and future prospects and explained those to investors. Having a well-respected analyst could help bankers land IPO mandates, and Gurley was one of the best. He’d already earned recognition from Institutional Investor, which polled investors to come up with analyst rankings.

Gurley was a native of Dickinson, Texas, a suburb roughly forty miles southeast of Houston’s city center and just nine miles from NASA’s Johnson Space Center, where his father, John, had been an early employee. Also named John, Gurley went by the shortened form of his middle name, William.

At Dickinson High School, Gurley distinguished himself as a very good student. Still known as John by some, he was one of just fourteen to graduate magna cum laude; one of many in the 259-strong class to win the Presidential Academic Fitness Award; and chosen for the Lamar University Scholarships.

He was also painfully aware of his height. A yearbook photo of magna cum laude graduates shows Gurley in the center of the last row, rising several inches over the next-tallest student. Gurley’s head is slightly angled. He played on the basketball team all four years.

His older sister went to work for Compaq, the early microcomputer market leader based down the road. She received options that paid off handsomely when the company went public. The younger brother was hooked on the thrill of technology, and credits that exposure to a realization that working in the technology industry could be both interesting and lucrative.

After high school, Gurley spent two years at Millsaps College in Mississippi, before transferring to the University of Florida to play basketball. Wearing No. 52, Gurley stood out not for his athletic prowess (he played as a reserve alongside future NBA players Dwayne Schintzius and Vernon Maxwell) but for his intellect. When the team’s depth came into question just before the NCAA tournament in 1988, his coach had this to say: “Bill and I both hope he doesn’t play. The only trips he’s made since early in the season have been to the science lab.” He actually did play—for one minute, during which he missed his only shot. Florida lost the game to Michigan in the second round. The Texan appeared on the Southeastern Conference’s Academic Honor Roll with a 3.77 grade point average in computer science.

After Florida, Gurley followed his sister to Compaq for a couple of years before he left to get his MBA from the University of Texas at Austin. By his own admission, Gurley entered UT as a mature student. He threw himself into studying financial concepts for a career on Wall Street. After graduation, he cold-called a number of well-known investment banks before getting a job at Credit Suisse First Boston (CSFB) in 1993.

It wasn’t long before Gurley made a splash for Quattrone’s DMG technology team. In early 1997, he downgraded Netscape Communications, the internet browser Hambrecht had helped take public. Microsoft was preparing a challenger, and Gurley worried about Netscape’s ability to weather the competition. The stock slumped nearly 20 percent on Gurley’s report. (The report so angered Marc Andreessen that he still holds a grudge.)

Gurley’s knowledge of the internet helped Quattrone’s team land the assignment to take a small online bookseller called Amazon.com public. Amazon.com was already a well-known name, and it was a surprise when founder and CEO Jeff Bezos selected Deutsche Morgan to lead the transaction. Many had expected Morgan Stanley or Goldman Sachs to underwrite the company’s IPO, but Bezos thought Gurley understood his business model better than most. The analyst spent hours talking to Bezos and began to put together his report.

When it came time to price the shares in the IPO, in the middle of May 1997, Gurley found himself in the New York lobby of the Four Seasons Hotel on the phone with Quattrone. The banker had been talking to Bezos, hashing out a price for Amazon.com’s shares.

Deutsche Morgan initially set a range of $12 to $14 for the shares, then increased the price to $16. Bezos thought he could get even more and wanted to raise it by another $2. Did Gurley have an objection? Quattrone wanted to know.

Gurley was comfortable with the higher price. Amazon.com priced its shares at $18 apiece.

The following day, Amazon.com opened near $30 before settling in for the close at $23.50. That represented a 31 percent pop, right in line with where corporate executives liked to see their stock after the first day of trading. Over the following days, Amazon’s stock declined, trading as low as $16.

Critics accused Bezos and DMG of pricing the deal too aggressively. They said no one would buy Amazon.com’s stock again after getting burned. The stock reversed its decline less than two months later and never again touched its IPO price as Amazon.com’s revenue and gross profit surged.

Bezos, Gurley, and Quattrone were proven right.

After less than two years working for Quattrone, Gurley was ready to try his hand at the venture capital industry. He lined up a job at Hummer Winblad Venture Partners, a respected firm based in San Francisco, and left Deutsche Morgan in late 1997.

The task of covering the internet had grown too large for any one person and Quattrone hired several analysts to replace Gurley. One of those was Lise Buyer, the T. Rowe Price analyst. Quattrone knew Buyer from his trips to Baltimore showcasing IPO candidates to T. Rowe Price’s portfolio managers.

Buyer was the most senior analyst Quattrone hired to cover consumer-facing internet companies, and she moved to California for the job. When Quattrone moved to Credit Suisse First Boston the next year, Buyer moved with him. By then, research analysts were widely quoted in industry and general-interest publications and well paid. Buyer was making more than $1 million a year.

She made the reverse move that Gurley and many others had before him, coming from the buy side and moving to the sell side. And she brought a fresh perspective. Knowing that investors often did the hard work of coming up with their ideas of future prices on potential stock picks, she refused to put price targets on many of her companies; at one point, she didn’t have targets on fifteen of the eighteen companies she covered. She felt that the industry was too new and the business prospects too uncertain for precision to be possible.

At Credit Suisse, a former colleague remembered, Buyer had strong opinions and wasn’t afraid to stick to them in the face of criticism. As a woman in a male-dominated field, Buyer defended her ideas and her territory, often in a frank, open, and straightforward way. When it came time to talk to investors, Buyer told them she had been in their seat and understood their position. She came across as analytical and confident.

As the internet boom continued in 1998, Buyer’s conservatism looked increasingly out of touch with how investors saw the industry. Or how company executives viewed the opportunity to sell shares and cash in on the enthusiasm. In July 1998, Mark Cuban’s Broadcast.com surged nearly 250 percent in its first day of trading. Readers of the following day’s New York Times were told that the stock had enjoyed “the best opening-day gain of any company in Wall Street history.” Over the following months, many investors and startup executives came to see first-day pops as a sign of an IPO’s success.

In September 1998, the online auction house eBay went public at $18 a share, with backing from Benchmark Capital, a venture capital firm set up in 1994 with the unusual model of sharing profits and investment decisions equally among a small group of partners. Benchmark had a more than 20 percent stake in eBay when it went public. The shares climbed 163 percent by the end of trading on the first day, a surge that annoyed eBay founder Pierre Omidyar because it meant that he could have sold shares at a higher price and collected more proceeds for his company.

Around that time, Credit Suisse bankers and Buyer spoke to theGlobe.com, an early social network, about a potential IPO. As with the rules Hambrecht had laid out to Bill Farinon in 1967, many investment banks still relied on revenue thresholds to guide their decision about a company’s readiness for the public markets. Credit Suisse’s analysts and bankers had their own threshold: if a company had less than something like $10 million in revenue, it was too immature for an IPO.

Buyer told theGlobe.com executives that they should wait a couple more quarters, till they had more revenue. Then they would look more like an IPO candidate. Bear Stearns, however, was much more cavalier. The bank, in the second tier of IPO underwriters, took theGlobe.com public in November 1998. The stock jumped 600 percent, the biggest one-day gain in history. Many other companies who had heard a similar message from Credit Suisse now flooded the phones and email inboxes of the firm’s bankers. Buyer lamented the degradation of underwriting standards when the analysts held their weekly call. Within weeks, Credit Suisse had joined most other investment banks in lowering their underwriting thresholds to match those of Bear Stearns. To resist meant losing out on IPO fees.

When CIBC Oppenheimer analyst Henry Blodget put a $400 price target on Amazon.com in December 1998, Buyer once again reminded readers of the Wall Street Journal how hard it was to value an internet stock. “It is not realistic to say that any of us can see out that far with any degree of accuracy.… No one should be confused that there’s any degree of certainty in a prediction of that kind.”

Blodget’s target was for one year. Amazon.com reached $400 within a month.

And Gurley joined Benchmark Capital, backers of eBay, in 1999.

By this time, Bill Hambrecht had become interested in exploring alternative approaches to bringing companies into the public markets. He set up a modest subsidiary of Hambrecht & Quist to experiment, seeking out smaller companies that were ignored by larger banks and open to new ideas. Hambrecht’s first couple of IPOs were successful, but it wasn’t long before his colleagues at Hambrecht & Quist clamored to take his deals for their larger, more established franchise.

Hambrecht’s desire to reform the IPO system was increasingly at odds with his own colleagues. One time, he allocated shares not to H&Q’s best trading clients but to those investors he felt would hold the stock for a long time. A couple of jilted investors put H&Q in the “penalty box,” steering their usual trading activity and its related commissions to other firms. “My sales department was ready to lynch me,” he recalled.

Hambrecht and CEO Daniel Case, the older brother of America Online CEO Steve Case, decided that it would be best if Hambrecht pursued his ideas on his own. In late 1997, the banker and venture capitalist announced his retirement from the firm he’d founded thirty years before.

The events of the mid-1990s had driven home at least three ideas that would propel Hambrecht into the next phase of his career.

• Price deals closer to the market clearing price so you don’t get as much of the pop that leads investors to flip shares.

• Get access to the customer base, because they are the most loyal shareholders a company will ever have.

• Find a way to avoid the conflicts of interest that had become so stark at firms that both underwrote IPOs and catered to trading clients.

Freed from the institutional structures of H&Q, Hambrecht set up his new firm, WR Hambrecht + Co., and set about designing something new. He had seen plenty of IPOs and thought that holding an auction for the shares might make sense.

Hambrecht set out to research auctions and got advice from professors at Stanford University who had designed an auction for the U.S. government to sell off access to broadband spectrum, which was becoming increasingly desirable with the rise of the internet. (Two Stanford faculty members, Robert Wilson and Paul Milgrom, won the 2020 Nobel Prize in economics for their auction studies.) Compared to broadband, a stock offering seemed much simpler, Hambrecht thought. “The old system made our firm very successful. My partners kept telling me don’t fix what ain’t broke,” Hambrecht said later. “But I couldn’t understand why you wouldn’t use an auction.”

He hired a technology whiz named Alan Katz from Texas Instruments to serve as his chief technology officer; Hambrecht & Quist had done a joint venture with TI. Hambrecht hired four developers to write the code that would become the OpenIPO system, an online auction marketplace designed to sell new shares more equitably.

Hambrecht made a couple of key decisions. The longtime banker knew enough not to try to cut out the large underwriters. He and his team designed the system so that they and their investors could participate in the auctions as well.

Hambrecht also based his technology on an algorithm designed by Canadian-American economist William Vickrey. Vickrey’s auction process involved collecting bids and then selling the asset being auctioned to the highest bidder but at the second-highest bid. It was intended to encourage bidders to submit the highest price they were willing to pay. The idea was that if investors knew that their bid was only used to win the auction, and that they would be paying the second-highest price, they would be more inclined to put in a higher bid. Vickrey was a Nobel Prize–winning Columbia University professor considered the godfather of congestion pricing for coming up with a way to balance the supply and demand of public transportation use.

Hambrecht’s system used a technique called a Dutch auction, named after the way flower merchants in the Netherlands sold their goods. The process involved offering securities at one price and then moving it lower until there were enough orders to sell the entire lot. It worked like this: If Hambrecht wanted to sell 10 million shares he might start the auction at $10. If investors only wanted to buy 2 million shares at that price, he would lower the price to $9. That might be enough to sell the other 8 milion shares, allowing Hambrecht to sell all 10 million. Those who put in a bid at or above $9 paid that price—regardless of whether their bid was higher, or their coziness with Wall Street. The process involved a new way to price and allocate shares but it still required Hambrecht’s firm to buy the shares from the company as an underwriter.

Hambrecht’s was the third such effort to try to sell new shares on the internet. Wit Capital, out of New York, was earlier to the idea. But Wit’s approach was drastically different from the OpenIPO system. It was geared to sell directly to retail investors. Started in 1996 by a former attorney at Cravath, Swaine & Moore to sell shares in the Spring Street Brewing Company, which made a wheat beer called Wit, the online brokerage partnered with other underwriters on dozens of deals. Wit Capital sold shares electronically to individual investors on a first-come, first-served basis. E-Trade, one of the first brokerages to trade shares online, also owned something called E*Offering.

Nonetheless, establishing the OpenIPO process took Hambrecht a year or so, with much of that spent negotiating with the SEC over the auction system’s rule set. Hambrecht, who had put $10 million into the company, found the regulators to be good partners. “They wanted to see it happen,” he said, “but we had to get everything approved by them.”

Once Hambrecht had the system ready to go, he sent out mailers intended to provoke his old investment banking colleagues and their potential clients. One side of the flyer read, “One for me. None for you. One for me. None for you.” The other side was more expansive: “Most investment bankers lead a very hard life. After they take care of each other they have to take care of their friends. And after they do that they have to take care of the people they’d like to have as friends. You can imagine that after all of this hard work there aren’t very many IPO shares to go around.”

Hambrecht announced the new business by giving a series of interviews. He was an insider now willing to disclose how the process worked. Hambrecht told Reuters that the OpenIPO would disrupt Wall Street’s allocation process and that this would mean a higher price for a company’s shares: “Most of the allocations go to the big institutions, because they go to a broker’s best clients. We’re opening up the process for groups currently excluded—small institutions and individual investors who have an affinity for the company.”

He added, “You get a more logical, rational price than the discounted one. The Dutch auction process ought to come out with a logical pricing that’s closer to where the real market level is going to be.”

In 1999, Hambrecht would get his first opportunity to try his new system. He signed up Ravenswood Winery Inc., a Sonoma County winery known for its zinfandels and an irreverent slogan: No Wimpy Wines. Hambrecht aimed to sell a million shares of Ravenswood in the range of $10.50 to $13.50. He charged the winery just 4 percent of the issue price, a healthy discount from the 6–7 percent that was then the typical underwriting fee for a traditional offering.

To bid in the OpenIPO system, investors had to have a brokerage account at Hambrecht’s firm or with one of five other small brokers who also signed up to participate in the process. There were other rules: Investors entered bids for how many shares they wanted to buy and at what price. After several weeks of collecting orders, Hambrecht’s process would set the price at whatever level was needed to sell all the shares. Anyone willing to pay more than that price would get all the shares they asked for, while those who bid at the price would get a percentage of their order filled.

No individual investor could get more than 10 percent of the entire deal, and the banker also kept the option open to limit sales to investors willing to purchase at least 1 percent of the entire issue.

In the end, Hambrecht was able to sell all one million shares, raising $10.5 million, though the final price was at the low end of the proposed range: $10.50. Shares rose 38 cents on the first day of trading.

It was a subdued opening to what Hambrecht hoped would be a game-changing proposition. About three thousand people put in orders, and roughly 75 percent of them got some shares. But Hambrecht also said he would stick to higher-profile companies in the future as he tried to get some traction for his new auction system.

In April 1999, the same month Hambrecht wrapped up his first attempt at reforming the system as an insider, a younger man started his first job in Silicon Valley as an outsider.

Barry McCarthy Jr. was in his late forties, a wiry former investment banker from New England with an itch to take a company public. He would soon get his chance. McCarthy had been on a ski trip with his family weeks earlier when he got a call from a recruiter looking to hire a CFO for an unproven DVD-by-mail business in California named Netflix. Netflix’s founders, Reed Hastings and Marc Randolph, needed a CFO who could withstand Hastings’s no-nonsense approach.

Hastings flew to meet McCarthy in Colorado, where the two men had dinner together. McCarthy soon accepted the position. His family stayed in Princeton, New Jersey—McCarthy had a child who was finishing school—and he drove west in his Ford Explorer listening to a series of tapes from the Practising Law Institute, a nonprofit that provided ongoing education for lawyers, about how to take a company through an IPO.

Hosted by Alan Austin, a partner at Wilson Sonsini Goodrich & Rosati, one of the first law firms to organize itself around technology clients, the lectures provided an overview of the SEC rules that governed public companies and their IPOs. McCarthy learned, for example, that the Securities Act of 1933 governs IPOs, while the Securities Exchange Act of 1934 governs companies once their stock is trading in the public markets. McCarthy had CFO experience, having worked at Music Choice for the previous six years, but no experience with public companies or IPOs.

Named after his father, an advertising executive, McCarthy grew up in Greenwich, Connecticut, with three siblings. The children’s grandfather had been mayor of San Francisco in the early years of the twentieth century. Barry Jr. attended Greenwich Country Day School before boarding at the coed Hill School, in Pottstown, Pennsylvania, where he played high school football, hockey, and lacrosse.

In 1971, McCarthy enrolled at Williams College, where he studied history and played lacrosse.

After Williams and the University of Pennsylvania’s Wharton School, McCarthy landed on Wall Street, on the trading floor at a firm, First Boston, that would become Credit Suisse First Boston. Tasked with keeping an eye on public policy issues and other research topics, he read popular newsletters that arrived by fax. One day, the organization governing accounting rules for public companies changed how companies could account for receivables. They could now sell loans and book the gain or loss immediately.

McCarthy noticed the implications almost instantly. The change meant that Ford could make loans to new car buyers and sell them right away instead of relying on a loan from a large bank. McCarthy got $40,000 from his boss to explore the potential of bundling the loans into securities. He hired a lawyer and a PhD to figure out the bond math. The small team borrowed the trust structure that had been created for mortgage-backed securities and called their creation receivables-backed certificates.

Infighting over who would own the new business ultimately prevented First Boston from using the structure. When others at the bank created something similar a few years later, few thought to credit McCarthy with the invention. But the project had a more personal impact. McCarthy discovered he wanted to run something himself, instead of advising others. He left the investment bank in 1990.

Even as he drove west into the glare of the first dot-com boom, McCarthy stuck to his East Coast ways. He arrived for his first day of work, on April 15, 1999, wearing a blazer.

At Netflix, McCarthy would develop an affinity for upsetting the status quo.