Profiling Snowflake — 47 APIs and the Data Cloud Control Plane

Snowflake is where a lot of the world’s analytical data actually lives, and the catalog shows 47 distinct APIs behind it. What’s striking is how few are about running a query. Most of Snowflake’s surface is a control plane — provisioning, governing, and operating the objects that the data sits inside.

What’s actually in the surface

The 47 APIs sort into clear planes:

The data plane — SQL API, Table, View, Dynamic Table, Schema, Database, Stage, Result. The actual read/write surface.
Compute — Warehouse, Compute Pools, Services, Image Repository. Snowflake separates storage from compute, and the compute control is its own set of APIs.
Data engineering — Task, Stream, Pipe, Dynamic Table, Function, Procedure, User Defined Function. The pipeline-and-transform machinery.
Governance — Role, Database Role, Grant, User, Network Policy, Managed Account. Access control as a programmable surface, which in a data platform is the whole ballgame.
Open formats — Iceberg Table, External Volume, Catalog Integration. The bridge to open table formats and external storage.
Cortex (AI) — Cortex Analyst, Cortex Inference, Cortex Search. LLM inference, natural-language querying, and retrieval sitting natively next to the data.
Operations — Alert, Event Table, Notification Integration, Notebook, API Integration.

What’s interesting about the shape

Governance is a large, first-class plane. Role, Grant, Network Policy, Database Role, and Managed Account are separate APIs because in a data cloud, who-can-touch-what is the product’s core promise. The contract reflects that.
Cortex put AI inside the data boundary. Rather than shipping data out to an external model, Cortex Inference and Cortex Analyst are APIs that run inference where the data already is. The AI surface is a peer of the SQL surface, not a bolt-on.
Open formats got their own contracts. Iceberg Table and External Volume are explicit acknowledgments that not all the data lives in Snowflake’s native format — and the API meets it where it is.

The takeaway

Forty-seven APIs is what a data platform looks like when storage, compute, governance, pipelines, and AI are each treated as independently operable. Snowflake’s surface is mostly control plane because that’s where the real work — and the real risk — lives. Walk it on the Snowflake provider page, and note how cleanly the AI layer slots in beside the data instead of around it.