Snowflakes fall from clouds. And in a multi-cloud world, one snowflake plans to rule them all. A $20bn snowflake is about to drop in the IPO market, creating its own thunderstorm, smack in the middle of a COVID-ridden summer.
How will this company change the cloud database landscape? What market forces are driving this company, which struggled mighty to raise its Series A round? As databases become great again, which companies are worthy of pursuit?
Data is powering every part of the economy. Snowflake is powering the data machines. For one, if Snowflake archives its true potential, it can become the next Oracle, even SalesForce of data.
As datalakes turn into dataswamps, Snowflake comes to the rescue: The deluge of data and falling storage costs have led us to become data hoards. We store and save everything — structured or unstructured. Every little byte goes into our gigantic and ever growing digital stomach and stays there forever. Till not too long ago, the world of databases was splitting into two — the cloud-native and legacy players. One boasted high reliability, low latency and speed — you could pass the ACID test (atomicity, consistency, isolation, durability) with flying colors. The cloud-native DB’s claim to fame was scale. Speed. Granted reliability was a bit weaker but it was designed for the new distributed age. As these two sides got into academic chest-thumping battles for superiority, the customers were stuck in a big messy problem — integrating various sources, managing the flow pipes was complex. The bigger pain at the executive level was ‘how do I get to business insights promptly and make data-driven decisions?” While getting the data into a data warehouse is not as difficult, staging and processing it, moving it to the analytics layer and delivering insights to various LOB managers is a painfully long journey. Taking weeks, if not months and prone to errors, it was a frustrating mess. Worse, with 5G and IoT data streams piping in faster and faster, the complexity for the C-Suite was growing exponentially.
You needed some adults to step into these data lakes, er….data swamps. Enter Benoit Dageville, co-founder of Snowflake, who had spent 15 years at Oracle, has a PhD and holds 80 patents. Benoit’s view was simple — how can he combine what he learned at Oracle and ride the cloud. “How could we create a new type of a database that would have the merits of SQL as well as the scalability of the cloud?” he wondered. With the ever expanding volume of specialized and operational workloads, datatypes and databases, it would take weeks, if not months to just get the data ready. The analytics would come later. It was hell for any data-driven CEO or LOB manager. “Just as the LOB managers now had their own apps, they were also demanding access to real-time data. In the current cloud native schema, that was not possible” says Rajkumar Sen, Co-founder of Blitzz.io, a cloud data replication company. “Snowflake solved a pressing problem and has truly abstracted the database, management and provisioning in a simple elegant way. And its business model is based on utilization — not subscription, which makes it even better for customers” he says. While ETL and ELT merely swap an alphabet, ELT is only possible at scale and offers much lower costs, thanks to the cloud. Extract, Transform, Load (ETL) — the legacy way was upended where you could Extract and Load data at the warehouse layer itself, eliminating the need for staging the data. This was seen as a step change and ELT is a welcome addition to the cloud native DB world.
When Snowflake was launched, Sutter Hill Ventures and Wing VC led the seed round at a ~$6m valuation in 2012. You’ve probably never heard of Mike Speiser — the first CEO of Snowflake. An entrepreneur and a VC at Sutter Hill, Mike gets the credit for this playing a seminal role at the start of this company. My friends on the street tell me that raising the Series A was a painful journey. Most VCs were afraid, asked obvious even inane questions like “What about AWS” or “Nah….Cloudera / Oracle will get on this quickly..they got this covered.” I know at least two partners who pushed hard to invest in Series A / B rounds but got turned down by the rest of their partnership. At its last round led by SalesForce Ventures, the company was valued at $12 billion. (Both partners have moved on and started their own venture firms).
The company was in stealth for a few years, had a new CEO and very few truly understood its potential. Once it got rolling, Snowflake has continued an unstoppable streak of revenue growth, customers, capital, at a pace that is the dream of every CEO and VC alike. As it’s ready to go public, it’s estimated to come out at a $20bn valuation. Guesstimates are that its revenues are in the $150 million to $200 million range, which grew 3X YoY. It now processes and estimated ~1 billion queries each day and boasts of over 2000 customers. Okta Business at Work 2020 reports put it as one of the fastest growing apps.
Snowflake is one of the fastest growing apps (Source: Okta Business at Work 2020 Report)
As the company scales, it’s eating away into AWS Redshift customer base rapidly. Its ability to offer multi-cloud / any-cloud capability allows it to remove any lock-in, cloud costs runoffs and create advantages for buyers.
The journey from $20bn to $100 bn….
Gartner states that DBMS (Database Management Systems) revenues has exceeded $55 billion in 2019, of which cloud DBMS is ~$17 billion growing 54%. It is projected to cross $100 billion by 2024. A fairly large market indeed, AWS has a clear lead with $9bn+. Frank Slootman, CEO of Snowflake has the background chops to claim a fair share of this $100 billion pie. When he was CEO of ServiceNow, a cloud based IT service platform, pundits claimed that the ‘cloud services’ market size was not too large, maybe a few billion. Slootman scaled ServiceNow revenues by a whopping 20X in 6 years, from $75 million to $1.5 billion (2017). In 2019, ServiceNow had $3.46 billion in revenues, growing 33% YoY and is setting its targets to $10 bn revenue mark. Its market cap is $84 billion and is a newly minted member of S&P 500.
Comparing Snowflake to ServiceNow, the “Data-Warehouse as a Service” market opportunity is at least 10X bigger. The tailwinds of multi-cloud, container and Kubernetes driven computational advantages are driving new demands for distributed storage. MongoDB, MemSQL, SAP and IBM have all seen ~100% YoY growth for their cloud DB business.
You know, we are living in a multi-cloud world…. 81% of respondents use 2 or more Public CSPs. (Source — Gartner, 2019)
Container storage and data management companies like Portworx and Databricks are riding this wave as well. The CEOs of these companies shared their views alongside Snowflake at a TechCrunch Disrupt event. Their observations were simple — new problems demand new thinking. Conventional approaches would not be able to handle the demands of scale, architecture and automation. And if COVID is the curse of the time, it surely has triggered digital transformation, much to the benefit of all cloud computing leaders. COVID has promptly cured all the laggards of any digital inertia.
But for Snowflake to continue to build its valuation curve, it cannot rely on Cloud DBMS alone. It has to integrate horizontally as well as vertically. It competes with, and has partnered with AWS, GCP and Azure — the big three Cloud Service Providers and has latched onto the AI wave with deeper and tighter integrations with partners like DataRobot and Databricks. Beyond AI, Snowflake will need to make the most of rising trends like distributed data mesh, which Oracle has already started to offer in its GoldenGate 19. Snowflake’s ability to lead the serverless opportunity may be a tougher challenge as FaunaDB is rapidly establishing its foothold. Backed by Madrona (an early investor in Snowflake) and Bob Muglia (former CEO of Snowflake) as its Chair, both know Snowflake’s playbook intimately. I’d speculate FaunaDB gets acquired by none other than Snowflake in due course. Other competitors like Dremio (which has raised over $100m from investors like Lightspeed, Redpoint and Cisco) will certainly keep the company on its toes. Fighting the big boys like Oracle, SAP, IBM and Teradata will not be easy — are not going to sit still, nor would AWS RedShift bleed silently. Finally, with open source eating away at the gross margins, new data analytics companies like Fishtown Analytics (backed by Andreessen-Horowitz and Amplify Partners) will continue to establish their presence on this vast landscape. Despite these various competitive threats, Frank Slootman has an indomitable spirit. Speaking at an annual CIO event, he once said that “Terror is a normal state for a CEO at all times” He knows storage — he started his career getting rid of those god-forsaken tape drive storage systems.
This is a cloud-based opportunity, a much better, bigger playground. Before taking on the role at Snowflake, he had retired from ServiceNow, sailing on a big boat. A classic founder’s move is to burn all boats and never turn back — Quemar Los Barcos — and Frank did just that. He sold off all his boats, jumped in the icy cold waters of boring middleware. His election motto may be “make DBMS great again”.
Slootman has set his sights on making Snowflake the single enduring platform that will democratize data — the oil of the digital economy needs to flow smoothly. The current pipes are leaden and old, filled with the sludge of DBA. Pipes cannot be repurposed quickly, they break and hungry customers / business unit ache for faster access to relevant data. “If Salesforce ushered in the SaaS world, we can do it for data” Slootman once remarked at a CIO Conference.
If he pulls it off, that’s the path to a $100 bn company.
(Mahendra Ramsinghani is an investor and author of two books “The Business of Venture Capital” and “Startup Boards” co-authored with Brad Feld. When he is not writing, he manages Secure Octane, investing in leading cloud infrastructure companies like Blitzz.io, Accurics and others. Like most VCs, he secretly wishes he had invested in Snowflake way back when….)