Snowflake's 111% Pop: The Other Way to Misprice an IPO

On 15 September 2020, Snowflake priced its IPO at $120 a share — above an already-raised range, implying a market capitalisation of about $33.3 billion.

The next morning it opened at $245, valuing the company near $68 billion before a single share had traded at the offer price on the open market. It closed just under $254, up roughly 111%, at about $70 billion — the largest software IPO ever.

This was reported almost universally as a triumph, and in one sense it obviously was. But it is worth stating the same event in different words: the company sold stock at $120 that buyers were willing to pay $254 for on the same day.

That is a pricing error. It is the exact opposite of the error Facebook made in 2012, and it is far less criticised — largely because the people who benefit from it are better represented in the coverage than the people who pay for it.

1. What Happened, in Numbers

• Range raised during marketing, then priced above the raised range at $120.
• Roughly 32.2 million shares sold once the underwriters' option was fully exercised, raising close to $3.9 billion.
• Implied market cap at pricing: about $33.3 billion.
• Opened at $245 — more than double the offer.
• Closed just under $254, a market value around $70 billion.
• Largest US listing since Uber's $8.1 billion offering in May 2019, and the largest software IPO on record.

Note what the raised range and above-range pricing tell you: the syndicate knew demand was strong and moved the price twice — and still ended up less than half of where the stock opened. This was not a case of nobody noticing. It is a case of a process that does not converge on the clearing price even when everyone involved is trying.

2. What "Money Left on the Table" Actually Means

The conventional calculation is the first-day gain multiplied by the shares sold. Here that is roughly $134 a share across about 32 million shares — north of $4 billion.

Two honest caveats before that number is used as an accusation.

First, the counterfactual is not clean. The company could not simply have priced at $254 and raised $8 billion. Some of the first-day demand existed because the deal was known to be underpriced — allocation itself was valuable, which drew in buyers who wanted the allocation rather than the company. Price at the clearing level and part of your book evaporates. The true recoverable amount is somewhere between zero and the headline, and nobody can tell you exactly where.

Second, the money did not vanish. It transferred. It went to the institutions and individuals who received allocation at $120 and could sell at $245 the same morning. The question is not whether value was destroyed — it was not — but whether the company's existing shareholders, including every employee holding equity, were the right party to fund that transfer.

Even at the conservative end, the sums are not marginal. Several billion dollars of primary capital is transformative for a company at that stage, and it was raised once.

3. Why This Keeps Happening

Underpricing is not an accident that recurs by chance for fifty years. It persists because the incentives produce it.

Allocation is a currency. An underwriter does not run one deal; it runs a continuous business allocating scarce shares to institutional clients it needs across the next decade of transactions. Shares that reliably trade up are a valuable thing to be able to hand out. Shares that reliably trade down are a liability. That asymmetry pushes the offer price down, and it does so without anyone behaving improperly — it is simply what a repeated game with the same counterparties produces.

The issuer transacts once; the bank transacts constantly. Your company will do this one time. Your bankers will price dozens more offerings for the same buyers. When your interests and theirs diverge, the relationship that persists is not the one with you. This is not a reason to distrust bankers — it is a reason to understand what you are buying, which our guide to how investment banks work sets out in detail.

A broken deal is career-ending; an underpriced deal is a headline. The penalty for pricing too high is severe and personal. The penalty for pricing too low is a congratulatory news cycle. Rational actors respond to that.

Fees are calculated on proceeds, which cuts the other way — a bank earns more on a larger raise. That is a genuine counterweight and it is why underpricing is a tendency rather than a rule. But the fee on the incremental amount is a few percent, while the value of a well-received allocation to a top client is measured in relationships. The counterweight is real and it is smaller.

4. The Defence of the Pop

The case for deliberately leaving something on the table is not frivolous, and a serious analysis has to engage with it.

• It builds a stable register. Buyers who make money on day one are more likely to hold, to buy more, and to participate in a follow-on offering.
• The follow-on argument. A company that lists well can return to the market later at a higher price. If you intend to raise again, a strong debut has genuine option value — though it is worth noting Snowflake did not need to return quickly.
• It is marketing. A doubling on day one is worth an enormous amount of attention in recruiting and sales, particularly for an enterprise software company selling to conservative buyers.
• Volatility protection. Nobody knows the clearing price in advance. A margin of safety prevents a broken deal, which is a far worse outcome for the company, its staff and its shareholders than a pop.

All of that is true. What it does not justify is 111%. There is a defensible band — conventionally somewhere in the region of a 10–25% first-day gain — that captures nearly all of these benefits. Beyond that, the incremental stability gained is small and the incremental transfer is enormous. The defence explains a pop; it does not explain a doubling.

5. The Detail Most Coverage Skipped

Alongside the offering, Snowflake sold shares through concurrent private placements, including to Berkshire Hathaway and Salesforce Ventures, at the IPO price.

This is worth understanding because it is now a common structure. A cornerstone or concurrent placement does several things at once: it de-risks the book, it lends the imprimatur of a respected buyer, and it locks in a holder who is not going to sell into the first week.

It also means a meaningful block of stock was sold at $120 to parties who were not subject to the same price discovery as the public book. In this instance the effect on the deal was constructive. The general point for a founder is that a cornerstone investor is buying two things — stock and a discount to the risk that the deal fails — and the second one is not free.

6. What the Alternatives Actually Do

The pricing problem has known structural responses, each with a real cost. Our comparison of IPOs, direct listings and SPACs covers the mechanics; here is what each does to this specific problem.

• Direct listing. No offer price at all — the opening auction discovers it. This eliminates underpricing by construction, because there is no discounted allocation to hand out. The cost is that you raise no primary capital in the traditional form and you get no book-building support, so it suits companies that do not need the money.
• Auction-format IPO. Buyers bid, the price clears where supply meets demand. Structurally sound and used successfully, but unpopular with syndicates for the obvious reason, and it removes the allocation control that some issuers genuinely value.
• Larger float. Underpricing is partly a function of scarcity. Selling more stock reduces the imbalance — at the cost of more dilution and more supply overhang.
• Pricing above the range, aggressively. Snowflake did exactly this and still missed by more than half, which is the strongest available evidence that incremental adjustment within the existing process is not sufficient.

7. Was $254 the Right Number Either?

It is tempting to treat the first-day close as the "true" value and the offer price as the error. That is too simple.

The opening price was set by a small fraction of the shares outstanding, in a market where most of the company was locked up, during a period of exceptional enthusiasm for enterprise software. A thin float with constrained supply does not produce a sober valuation; it produces the price at which the most optimistic available buyer clears. That first-day close proved not to be a durable level, and the stock traded below it for extended stretches in the years that followed.

So the honest framing is not "the bankers underpriced by $134." It is that the offer price and the first-day price were both artefacts of process rather than measurements of value — one set by a negotiated book, the other by a supply-constrained auction. The company's realised proceeds were determined by the first. Everything else was commentary.

This is the same lesson the WeWork filing taught from the other end: a number produced by a constrained process is not a valuation, whichever direction it points.

8. The Lessons That Transfer

• Judge the pop, not the direction. A modest rise is a well-run deal. A doubling and a break are both process failures; only one of them gets applauded.
• Ask your bankers to model the transfer explicitly. At the proposed price, what is the expected first-day gain, and what does that cost your shareholders in dollars? Requiring the number in writing changes the conversation.
• Understand who is in your book. Allocation decisions determine who owns you and how long they intend to. Ask for the allocation policy before pricing, not after.
• Float size is the lever you actually control. A larger float reduces both the scarcity premium and the later lockup shock.
• If you do not need the capital, question the structure. Direct listings exist for exactly that case, and the companies that have used them were mostly not short of money.
• Employees pay for this too. Dilution from an underpriced primary raise falls on everyone holding equity, and staff are rarely represented in the pricing conversation.

Frequently Asked Questions

Is a big pop actually bad for the company?

It is bad for the balance sheet and good for the narrative, and reasonable people weigh those differently. What is not defensible is treating it as unambiguously positive. The company raised roughly $3.9 billion where the same shares were worth over $8 billion to the market that morning. Whether the resulting attention, register stability and follow-on optionality were worth that difference is a real question — but it is a question, not a settled point.

Why do the banks not just price higher?

Because they are pricing a book of repeat clients rather than a single transaction, because a broken deal is punished far more harshly than an underpriced one, and because nobody actually knows the clearing price in advance. The incentives lean one way and the uncertainty gives cover. That combination is durable, which is why underpricing has persisted through decades of criticism.

Could Snowflake have done a direct listing?

Plausibly — it was a well-known company with a strong balance sheet, which is the profile that suits the structure. The trade-off is that a direct listing would not have raised primary capital in the same way, and the concurrent cornerstone placements that added credibility would have needed a different structure. It is a genuine alternative, not an obvious one.

Does a large pop predict long-term performance?

No, and the belief that it does is one of the more expensive retail misconceptions. First-day performance is a function of allocation scarcity and float size, not of business quality. Some heavily popped listings have compounded for years; others gave the entire gain back and more. Treating the pop as a quality signal is confusing the process with the company.

What should a founder actually do with this?

Decide in advance what you are optimising for — proceeds, register quality, or attention — because those pull in different directions and the default process quietly picks for you. Then make the pricing trade-off explicit with your board rather than delegating it, and treat the board's role here as substantive rather than a rubber stamp on the night.

Does this apply to the very large listings happening now?

Directly. The current cohort is larger, later-stage and thinner-floated than the 2020 class, which sharpens both failure modes at once — a scarce float invites a violent debut, and a very large raise makes any percentage of underpricing an enormous absolute number. Our coverage of the SpaceX listing and the AI pipeline looks at how those tensions are being handled at unprecedented scale.

The Bottom Line

Snowflake's listing was a commercial success and a pricing failure at the same time, and the market only has vocabulary for the first.

Facebook priced to capture every dollar and had no aftermarket. Snowflake priced to guarantee an aftermarket and gave away billions. The defensible answer sits between them, and getting there requires a founder who treats the offer price as a decision to be argued rather than a number delivered on the night before trading.

Global Capital Network connects founders, investors and the bankers and advisors who price these deals. See upcoming events or get in touch.

Accurate as at 2 August 2026. Figures are as reported contemporaneously. Primary sources: CNBC on pricing, CNBC on the debut, and company filings via SEC EDGAR. This article is general information and market history, not investment advice.