Snowflake's IPO left almost $5 billion on the table. Maybe it wasn't so successful after all

No recent event better exemplifies the tech frenzy than the jaw-dropping Snowflake IPO. Following its debut on Sept. 16, many in the media marveled at how its $70 billion valuation made the software unicorn––its annual sales running at just $500 million—the priciest newly listed player ever. What went mainly unnoticed is that Snowflake captured another distinction as the second-most underpriced IPO in history, as measured by dollars “left on the table,” and the record holder, by a wide margin, for a newcomer of its slender size. “We’ve seen an unusually high number of IPOs this year that rank near the top in forgone proceeds,” says Jay Ritter, a professor at the University of Florida and the nation’s leading IPO expert. “But Snowflake dwarfs them all.”

Prior to the offering, Snowflake’s team of two dozen underwriters, led by Goldman Sachs and Morgan Stanley, sold 28 million newly issued shares to their institutional clients at a price of $120. That part of the sale raised $3.36 billion, less commissions of $106 million, netting $3.254 billion for Snowflake’s treasury. But it also garnered additional cash by selling an extra 2.083 million shares each to Warren Buffett’s Berkshire Hathaway and cloud-startup investor Salesforce Ventures at the same price. Technically, the 4.17 million share deals qualified as a separate secondary offering following the IPO.

The bankers also reserved the right to boost the allocation for their customers by 4.2 million shares, via an “overallotment” provision known as the “greenshoe.” As it turned out, demand was so strong that the underwriters found eager buyers among their clients for the greenshoe shares as well. That brought the full number sold to around 36.4 million. After fees, the three transactions brought Snowflake $4.364 billion in cash.

As often happens with glamour IPOs, investors willing to pay a lot more than $120 a share either couldn’t get any shares or obtained only a fraction of what they wanted. When Snowflake debuted, they arrived en masse, pushing the price up 112% to $253.93 by the opening day close. Had Snowflake sold those 36.4 million shares via a direct listing, or an auction process that captured the shares’ full value, it would have amassed, even after the same $106 million in fees, $9.24 billion. So the eight-year-old phenom in effect sacrificed $4.88 billion. In other words, it “paid” $1.12 in forgone funds for every dollar it raised.

Snowflake CEO Frank Slootman defended the pricing in an interview with Fortune on Wednesday. The company wanted to sell IPO shares to large and steady institutional shareholders who wouldn’t pay the highest price, he said, while deriding the later trading action as the “frothy, opportunistic retail side.”

That $4.88 billion is by far the largest forgone amount in a dozen years, and ranks second to Visa’s underpricing of $5.85 billion at its entrance in 2008. This year has featured many of the biggest first-day bounces ever for IPOs. No fewer than 13 of the 60 biggest amounts ever left on the sidelines came in the first eight-and-a-half months of 2020. Yet all the other spikes pale beside the Snowflake pop. Snowflake’s first-day shortfall was three times as large as that logged by No. 2, Corvis, and an amount equal to the count for the next four combined, Royalty Pharma, ZoomInfo, Vroom, and nCino.

Of course, forgoing $4.88 billion may not seem too much of a loss when the investors reckon your company is worth 15 times that number or, as we’ll see, even more. Still, Snowflake is a young, fast-sprinting player that’s now burning over $200 million a year in cash, and will need increasing amounts of capital to keep racing. Having an extra almost-$5 billion in its coffers would provide plenty of additional fuel for future expansion. Piling all that extra cash on its balance sheet would also have raised its book value, and hence its valuation, by a not inconsiderable near 7%.

In reality, investors on opening day were valuing Snowflake at a lot more than $70 billion. That’s because its total shares outstanding will grow substantially in the years ahead as employees cash in their generous equity grants. Its prospectus discloses that Snowflake has awarded 72.2 million options and another 7.7 million shares of restricted stock, or roughly 80 million in shares poised to swell the float. The average exercise price on the options is a paltry $6.70, and the company has met the targets required for the restricted stock to vest. In fact, the prospectus discloses that 24 million of the options have already vested.

“In most cases, since options vest over several years and employees often leave before the options and restricted shares vest, many of the shares expire and aren’t added to the float,” says Ritter. “But in the case of Snowflake, the options are so far in the money that it’s the ultimate in golden handcuffs. Who would want to leave and sacrifice shares now selling at 40 times the strike price?”

Right now, it looks like almost all of those 80 million shares-in-waiting will add to Snowflake’s current float of around 280 million, lifting the total by 29% to 360 million. So that looming dilution means Snowflake is sporting—at the Sept. 16 price of around $250—what we’ll call a shadow valuation of more like $90 billion.

That’s the number that counts. The $90 billion bogey will dictate what Snowflake must earn going forward to provide the big returns investors are expecting. In these heady times, forgoing almost $5 billion may seem a small price to pay for notching an IPO that both succeeded in raising billions and brought a rousing vote of confidence by bestowing one of the richest valuations in tech history. But if times turn stormy, Snowflake’s leaders could rue letting all that rainy-day money get away.