Snowflake

Snowflake's thesis split in two during 2024. The original bull case — an analytics-only data warehouse with 30% growth and 75% gross margins — broke when the post-Covid hyperscaler optimization wave hit. Net revenue retention fell from a peak of 178% in 2022 to roughly 125% by 2025, and the stock dropped over 60% from its all-time high. The 2026 story is the rebuild under Sridhar Ramaswamy, the former Neeva CEO and ex-Google senior VP who took over from Frank Slootman in February 2024. Ramaswamy's thesis is that Snowflake stops being «the warehouse» and becomes «the AI data cloud» — meaning customers run not just SQL analytics but generative-AI workloads, vector search, and ML training directly against their Snowflake-resident data via Cortex AI. The early signal is genuinely encouraging: Cortex revenue crossed $200 million annualized run-rate exiting Q1 2026, and the Snowflake Marketplace AI app category grew 4× year-over-year. Total FY2026 product revenue guidance is $4.85 to $4.92 billion, implying 24% growth at the midpoint — not the 30%+ era but a believable re-acceleration.

Smart-money positioning on Snowflake is bifurcated. Bulls include long-tenure software specialists Iconiq Capital and Whale Rock Capital, both of whom added meaningfully through the 2024-2025 drawdown. Brad Gerstner's Altimeter Capital held a top-10 position into late 2025. The activist non-presence is itself the tell: no major activist has built a position, suggesting both that the strategic playbook is intact and that there is no balance-sheet lever to pull. On the other side, the persistent short interest near 5% reflects hedge funds betting that Databricks (still private) and Microsoft Fabric will compress Snowflake's gross margin and growth simultaneously over the next 24 months. Berkshire Hathaway sold its full Snowflake position in 2024 — a meaningful absence given how visibly Munger had championed it pre-IPO.

Snowflake trades at 11.2× forward EV/revenue and 47× forward FCF — both materially below the IPO-era valuations of 50× revenue but a notable premium to mature SaaS at 6-8× revenue. The premium is now justified by demonstrable 25% growth plus expanding free cash flow rather than the speculative 50% growth narrative of 2021. Bull case (Cortex drives NRR to 135%, Iceberg captures new enterprise tier, SBC contained): fair value $235. Base case (steady 24% growth, gross margin holds 75%): $185. Bear case (Databricks wins enterprise, NRR slides to 115%): $115.

1. August 27, 2026: Q2/FY2026 earnings — first quarter where Cortex AI workloads are individually broken out; watch for net new customer count vs Databricks reference signals.
2. November 2026: Snowflake Summit Europe — major product announcements expected, particularly around vector search and agentic-AI orchestration.
3. Throughout 2026: Databricks public-listing watch — an S-1 filing or IPO would create direct comparable-pricing pressure but could also rebase sector multiples upward.

I am long Snowflake at roughly 2% of my portfolio. The Ramaswamy transition was the moment to reconsider this name: he brings a credible AI background and unlike many product-led CEOs, he is articulate on the unit economics. The Cortex run-rate growth is the single most important number to watch — if it triples to $600M+ by Q4 2026, the multiple expansion is inevitable. The bear case is real but largely priced in at current levels. Where I think consensus is wrong: most analysts still model Cortex as a margin headwind rather than a revenue accelerator. The 25%+ free-cash-flow margin is the safety net — if the AI thesis works only partially, you still own a profitable software business growing 20%+ with 5% annual buyback yield. That is a much better setup than 2022-2023 Snowflake.