Snowflake's Debut Was the Biggest Software IPO Ever
The data warehousing company priced at 120 dollars, opened at 245, and closed its first day near 254. The gap between the price the company received and the price the market paid is the story.
The Numbers
Snowflake priced its initial public offering at 120 dollars per share on September 15, 2020. The stock opened the following day at 245 dollars and closed its first session near 254 dollars, roughly 112 percent above the offer price. It was the largest software IPO ever completed.
The headline framing was triumphant. A closer look at those three numbers describes a substantial transfer of value away from the company.
How an IPO Actually Prices
In a traditional underwritten IPO, investment banks build a book of demand from institutional investors, then set an offer price. Shares are allocated to those institutions at that price. The company receives the offer price multiplied by shares sold, minus underwriting fees. That is the entire proceeds to the company.
Whatever happens after trading begins does not send another dollar to the company. If the stock doubles on day one, that gain belongs to whoever received an allocation at the offer price, not to the business.
The company sold shares at 120 dollars that the market immediately valued at 245. Every dollar of that gap was capital the company could have raised and did not.
The Cost of the Pop
Multiply the gap by shares sold and the figure is enormous, well over a billion dollars of value that went to allocated investors rather than onto Snowflake's balance sheet. Underwriting fees, typically a few percent, are trivial next to a pricing gap of that size.
Defenders of the practice make two arguments worth stating fairly. A strong debut generates attention and can help recruiting and customer confidence. And underpricing compensates institutional investors for taking risk on a company with no public trading history, which supports demand in future offerings. Both have some merit.
The counterargument is simply the arithmetic. A company gets one first sale of its shares, and leaving that much on the table is a real cost borne by existing shareholders through dilution. Founders and employees own a smaller slice of the company than they needed to.
Why Alternatives Exist
Dissatisfaction with this dynamic pushed companies toward two alternatives during this period. A direct listing places existing shares on an exchange without issuing new ones, letting the market set the opening price with no allocation process. It raises no new capital in its classic form, so it suits companies that do not need cash.
A Dutch auction, used most famously by Google, sets the price where demand actually clears rather than where bankers judge it should sit. Both approaches attempt to narrow the gap between the company's proceeds and the market's valuation, and neither fully displaced the traditional model.
The Business Underneath
The enthusiasm was not baseless. Snowflake sold cloud data warehousing on a consumption model, charging for storage and compute actually used rather than a fixed seat license. That model produces net revenue retention above 100 percent when customers expand usage, meaning revenue grows from existing accounts before any new customer is added.
It was also unprofitable and priced at an extraordinary multiple of revenue. Berkshire Hathaway's participation drew attention precisely because it looked so unlike a typical Buffett investment, and the deal was reportedly driven by a different manager within the firm.
The Bottom Line
Snowflake's debut set a record and cost the company more than a billion dollars of forgone capital. A first day pop is a headline for the press and a bill for the issuer.