The AI Data Cloud that made the data warehouse a multi-cloud service
HQ: Menlo Park, CA, United States (Montana 'no-headquarters' registration) | Founded: 2012 | Employees: ~9,000–12,000 (SEC 2026 proxy cites ~9,060; some datasets report ~12,475) | Stage: Public — NYSE: SNOW, IPO September 16, 2020 | Website: https://www.snowflake.com
Snowflake is a cloud-based AI data platform company that delivers a single SQL engine, governed data cloud and increasingly an AI runtime across Amazon Web Services, Microsoft Azure and Google Cloud. Unlike legacy data warehouses licensed per customer per cloud, Snowflake meters storage and compute separately and lets customers share live data sets across accounts and companies through secure data sharing and the Snowflake Marketplace.
The company was founded on July 23, 2012 by Benoît Dageville and Thierry Cruanes, data architects with deep parallel-execution experience at Oracle, and Marcin Żukowski, co-founder of Vectorwise. Sutter Hill Ventures' Mike Speiser was first CEO; Bob Muglia (ex-Microsoft) took over in 2014 and brought the product out of stealth with ~80 design customers. Frank Slootman (ServiceNow, Data Domain) became CEO in 2018, ran the 2020 IPO, and remains chairman. Sridhar Ramaswamy, co-founder of search startup Neeva, became CEO in February 2024 and has reoriented the roadmap around Cortex, coding agents and enterprise AI.
Today Snowflake is used by roughly 790–813 of the Forbes Global 2000 for analytics, data engineering, data sharing and, increasingly, LLM inference and agentic workflows against governed enterprise data. The platform's signature economics — high concurrency, per-second compute billing and zero-copy sharing — made it the default cloud data warehouse of the 2020s and position it as a primary candidate to be the AI data cloud of the late 2020s.
Snowflake's core insight was architectural: data warehouses had been bolted onto single clouds and single customers, but the value of data multiplies when it can be shared. Separating storage from compute, and running the same SQL engine across AWS, Azure and GCP, let Snowflake sell one product to multi-cloud enterprises that previously had to pick a hyperscaler. That design choice is why Capital One, BlackRock, Siemens and Thomson Reuters run on it, and why net revenue retention has stayed above 120% for years — customers keep putting more data in and turning on more compute.
The fiscal 2026 and early fiscal 2027 numbers show a company that has re-accelerated after the 2022–2023 growth scare. Product revenue of $4.47 billion in FY26 (+29%) and $1.55 billion in Q2 FY27 (+35%) with RPO near $9.8 billion means the demand pipeline is expanding faster than revenue recognition. Management shipped 430+ product capabilities in FY26, including Cortex Code and Snowflake Intelligence, and acquired Observe to enter the $50B+ IT operations market.
The risks are familiar for a platform at this scale. Databricks remains the most credible rival, blurring the lakehouse and data-warehouse line with Spark and open table formats. Hyperscalers keep improving BigQuery, Redshift and Synapse, and some of Snowflake's largest customers also operate their own AI model stacks. The business is still consumption-based, so macro-driven dips in query volume hit revenue immediately. And while Cortex is the right strategic bet, it has to out-execute both hyperscaler-native AI tools and open-source model ecosystems running on the same data.
Browse: Home | Blog | About | All Startups A-Z | View full profile on StartupWiki