Snowflake’s IPO is a bet on companies using AI for everything

Snowflake’s IPO caused a flurry of interest in cloud data.
Snowflake’s IPO caused a flurry of interest in cloud data.
Image: REUTERS/Navesh Chitrakar
We may earn a commission from links on this page.

Snowflake, the buzzy cloud computing company, just delivered the biggest software IPO ever by betting on a future in which all businesses increasingly rely on big data and AI to make decisions. Its stock began trading on Sept. 16 at $245 per share—more than double the $120 price Snowflake set the day earlier—giving it an opening valuation of $67.9 billion.

The startup’s main value proposition is that it makes it easier and cheaper for businesses to analyze data they’ve shelved away on the cloud—including the massive datasets needed to train machine learning algorithms. Because Snowflake structures its software differently, it can run resource-intensive AI programs more efficiently than its competitors, including juggernauts like Amazon’s AWS and Microsoft’s Azure.

“Snowflake is very important because they can challenge the Amazon AWS power,” said Per Roman, managing partner and co-founder of investment firm GP Bullhound. Amazon’s cloud service controls nearly half the global market, and takes much of the web down with it whenever it experiences disruptions.

Snowflake is only able to take on a dominant, entrenched competitor like AWS because it has come up with a better solution for one problem: allowing companies to cheaply expand their reliance on AI to help them with an ever-growing set of business decisions. That is the vision Wall Street valued at roughly $70 billion today.

AI requires two key ingredients: huge troves of data and a lot of computing power. If you have both, you can monitor the performance of your business in granular detail (and in real time) to predict future conditions. Snowflake can help companies do this kind of analysis more often and at a lower cost: In July, Snowflake handled 507 million requests per day from companies looking to track key data points in real time, according to its IPO filing.

Snowflake differentiates itself from competitors like AWS through its software architecture. The company divides its massive pool of computing power into three groups: One is dedicated to storing data, another is just for analyzing that data, and the third is a brain that keeps the other two running. The system can dedicate more resources to storage or data analysis on the fly, depending on demand.

AWS offers similar services, doesn’t separate data storage and analysis, which means clients effectively pay for more analytical computing power they may not use. Snowflake, by comparison, only charges clients for exactly as much storage and computing power as they need. Snowflake also makes it easier for companies to share data with clients and partners, or to buy and sell datasets on an internal marketplace.

Roman said that every company is trending towards a future in which they’ll rely on cloud services to store and analyze data. “All enterprises, from the largest in the world to your small mom-and-pop shop, need to migrate their computing storage and systems into the cloud,” he said. Snowflake’s soaring valuation suggests that investors have bought into this bet on ubiquitous AI.

But as businesses are increasingly adopting AI to inform their strategies and automate everything from resume review to ER triage, a string of high-profile scandals has made corporate executives think twice about their use of big data. Shifting public opinion, new government regulations, and the cost of compliance could change companies’ calculus about AI—and send Snowflake’s valuation drifting back down to Earth.