Three database engineers separated storage from compute — Snowflake built a cloud giant on the idea

Snowflake was designed around a question that traditional data warehouses struggled to answer: why should storing data and processing it have to scale together?

The company was founded in 2012 by database engineers Benoît Dageville, Thierry Cruanes and Marcin Żukowski. Dageville and Cruanes had deep experience at Oracle, while Żukowski brought database and analytical-engine expertise.

They decided to build the new platform directly on public-cloud infrastructure rather than adapt an older on-premise database for hosting.

Separating storage and compute changed the economics

Traditional analytical systems often tie storage capacity closely to the computing hardware running queries.

Snowflake separated those layers so customers could store data independently while creating different compute clusters for different workloads.

A finance team, data engineer and machine-learning workload could therefore use the same underlying data without having to compete for one fixed pool of processing power.

The public cloud was treated as an ingredient, not merely a hosting location

Snowflake’s founders designed the service to use elastic cloud infrastructure from the start.

This made automatic scaling and consumption-based use part of the architecture rather than a retrofit.

The company later expanded across multiple public-cloud providers, allowing customers to run Snowflake in different regions and clouds while keeping a similar operating model.

A data warehouse gradually became a broader data platform

The first clear use case was cloud data warehousing: centralising structured data for analytics.

Snowflake later expanded support for semi-structured data, data lakes, data sharing, application development and machine-learning workloads.

The product vocabulary changed with the expansion, moving from cloud data warehouse to Data Cloud and later AI Data Cloud.

Data sharing turned duplication into a platform feature

Organisations often exchange information by exporting files and creating copies.

Snowflake developed sharing mechanisms that allow authorised users to access governed data without requiring the same pattern of manual copying.

This became the foundation for marketplace and collaboration features connecting data providers with customers.

The 2020 IPO became one of software’s biggest market debuts

Snowflake began trading on the New York Stock Exchange under the symbol SNOW in September 2020.

The IPO arrived during rapid enterprise migration towards cloud infrastructure and gave Snowflake enormous visibility as a pure-play data platform.

Frank Slootman scaled the business, then handed it to an AI technologist

Frank Slootman became chief executive before the IPO and led Snowflake through its public-market expansion.

In February 2024, he stepped down as CEO while remaining chairman. Sridhar Ramaswamy, previously Snowflake’s senior vice president of AI and a former Google advertising executive, became chief executive.

The succession made the company’s AI ambitions unusually explicit.

Generative AI made governed enterprise data more valuable

Large language models can generate text, code and analysis, but enterprise deployments depend on connecting those models to trusted internal information.

Snowflake has increasingly built AI and application-development capabilities alongside its data platform so customers can run models closer to governed data.

The business opportunity is not simply model training. It is making corporate data usable by AI without creating another uncontrolled copy of that data.

The original architecture still explains the modern company

Snowflake has accumulated many products since 2012, but its founding design decision remains central.

Separating storage, compute and cloud services created the elasticity that allowed the platform to expand from data warehousing into sharing, applications and AI.

For TechnologyBlog.co.za, this history page can anchor future Snowflake database, data-platform and enterprise-AI coverage while keeping product articles focused on what has changed now.

Primary sources reviewed include Snowflake’s founder-authored architecture material, official company information, its 2020 IPO release and the 2024 CEO transition announcement. Information reflects public material checked on 18 September 2026.