Deep Research

Snowflake (SNOW)

3 min read

Snapshot

  • Ticker: SNOW (NYSE)
  • Bucket: Software in the AI Crosshairs
  • Q1 2026 position: 144,818 shares, 100M+
  • HQ: Bozeman, Montana (principal executive office; officially “headquarterless”)
  • What it does: Cloud data platform — warehousing, data engineering, data sharing, and an expanding AI layer (Cortex) sold on consumption pricing across AWS/Azure/GCP.

Business Overview

Snowflake remains one of the strongest growers at scale in software. Fiscal 2026 (ended Jan 31, 2026) product revenue was 9.77B (+42%). Fiscal Q1 2027 (reported May 27, 2026) showed acceleration: product revenue of 1M of trailing product revenue (+29%), and a raised FY27 guide of $5.84B product revenue (+31%) at a 13.5% non-GAAP operating margin. Under CEO Sridhar Ramaswamy the company has pushed hard into AI workloads — Cortex LLM functions, AI SQL, interoperability via Iceberg open table formats — to position the platform as the data substrate for enterprise agents.

Why Atreides Owns It

Snowflake sits squarely in Baker’s “Software in the AI Crosshairs” frame: SaaS and consumption-software vendors face a “life or death decision” about sacrificing margin to ship AI, while “anything you can verify, you can automate.” For years Atreides treated SNOW as a core data-layer holding — $100M+ continuously from Q4 2024 through Q4 2025 — on the logic that whoever owns governed enterprise data taxes every AI application built on it.

The Q1 2026 trim of ~70% (469K → 145K shares) is therefore a thesis statement, not noise. It happened in the same quarter the fund exited GitLab entirely and rotated software exposure toward runtime/infrastructure names (Palo Alto, Akamai, Zoom, Twilio). Plausible drivers, unconfirmed: Databricks’ and the hyperscalers’ converging lakehouse/AI stacks compressing Snowflake’s differentiation; open table formats (Iceberg) loosening data gravity; and valuation — SNOW re-rated sharply through 2025, and Atreides habitually sells what has worked. The residual stub keeps the name on the sheet without underwriting the next leg.

Position History

QuarterTypeShares/NotionalValue% of 13F
Q4 2024Common756,767$116,852,3922.56%
Q1 2025Common608,715$88,969,7842.70%
Q2 2025Common525,769$117,651,3293.26%
Q3 2025Common469,545$105,905,8752.06%
Q4 2025Common478,754$105,019,4771.28%
Q1 2026Common144,818$21,841,4510.44%

Five quarters of patient, roughly 105M even as % of 13F fell (the fund grew), so the Q1 2026 cut reads as an active de-rating of the thesis rather than portfolio rebalancing.

Risks

  • The crosshairs problem: hyperscalers and Databricks bundle competing data/AI stacks at marginal cost; open formats erode switching costs.
  • Consumption-model sensitivity: revenue tracks customer compute optimization cycles; AI workloads are upside but also rebate leverage for buyers.
  • Margin-versus-AI tradeoff: shipping Cortex/agents requires GPU spend that pressures the newly positive operating margin — exactly Baker’s SaaS margin-squeeze scenario.
  • Valuation: ~30% growth is already paid for at premium multiples; deceleration below guide would compound multiple and estimate cuts.
  • Signal risk for followers: the manager who held this for years just sold 70%; the remaining position is closer to an exit ramp than a conviction holding.

Sources

Backlinks

Graph View