Snowflake runs a data platform on top of the major clouds: companies load their data, query it with SQL, share it with partners and increasingly build AI and machine-learning workloads on it. Customers pay for the compute they consume and the storage they use, not seats.

Revenue in FY2026 (year ending January 2026) was $4.7B, up 29.2%.

The original advantage was architecture: separating storage from compute so each scales independently, with near-zero administration. Data sharing between Snowflake accounts creates network effects within industries.

The moat is narrowing by design. Open table formats (Apache Iceberg) let customers keep data in their own storage and query it with any engine, including Snowflake's main competitor Databricks and the clouds' own services. Snowflake has embraced Iceberg, which keeps customers but makes leaving easier.

Gross margin near 67% (it pays the clouds for the compute underneath). Free cash flow was $1.1B in FY2026, but stock pay was $1.6B, a third of revenue, so owner earnings were about -$0.5B and the GAAP operating margin was -31%. Buybacks ($0.9B in FY2026) slow but do not stop the dilution: shares still rose 1.4%.

Data platforms are central to AI, and Snowflake's consumption model grows with usage. But the economics belong to employees as much as owners today, and open formats cap pricing power.

We are watching; we would need to see stock pay fall sharply as a share of revenue.

• Stock pay remaining near a third of revenue.
• Open formats and Databricks compressing what customers pay for compute.
• Consumption volatility when customers optimise.

• Owner earnings turning positive (a reason to look again).
• Product revenue growth falling below 20%.
• Large customers moving workloads to open engines on their own storage.