Five Things the Market Is Missing About Snowflake's Horizon Catalog Pivot

Enterprise AI adoption is creating a category that did not exist three years ago: provenance, lineage, and policy enforcement on the data that feeds AI workloads. The compliance overhead is substantial; CISOs at Fortune 500 customers are writing data governance policies for AI-specific use cases that did not need policies in the pre-genAI era.

Snowflake's Horizon Catalog positioning targets that compliance need directly. The product is integrated with the Snowflake data plane, which means the governance metadata travels with the data without requiring a separate ETL pipeline. That is a meaningful architectural advantage over competitors trying to layer governance on top of data lakes that they do not natively control.

The historical parallel is the early evolution of identity and access management as a category. Once IAM became embedded in the cloud platform layer (Active Directory, IAM in AWS), the switching cost on the underlying platform increased materially. Horizon Catalog plays the same role for AI data governance.

Snowflake's net revenue retention bottomed at roughly 127% in late FY24 after running near 165% during the 2022 cycle peak. The most recent quarterly print delivered 130%, the first sequential improvement after seven quarters of compression.

This is the data point that the buy-side consensus has been waiting for. The pattern in software franchises is that NRR inflection precedes revenue growth reacceleration by roughly two to three quarters. Snowflake's NRR turned in calendar Q4 2025; the revenue growth reacceleration should be visible in mid-2026 quarterly prints.

The AI workload contribution to consumption is the operative driver. Snowpark, Cortex, and the broader AI/ML compute layer are attracting incremental consumption that was not part of the franchise's growth profile two years ago. The CFO commentary on the most recent earnings call referenced AI-related consumption hitting double-digit percent of new growth, up from negligible levels 18 months ago.

Databricks remained the primary competitive threat through 2023-2024. The competitive geography has shifted in 2025 in two ways. First, Databricks' delayed IPO has reduced its ability to execute aggressive customer acquisition pricing, since the cash burn discipline required for an eventual public listing limits the discounting flexibility.

Second, Microsoft Fabric, which was the secondary competitive threat in early 2024, has had a slower commercial uptake than the launch hype implied. Customer surveys we track show Fabric mentioned in roughly 12-15% of new RFPs, well below the 25%+ the launch coverage suggested. Snowflake remains the default for new enterprise data warehouse decisions.

The smaller competitors (Dremio, Starburst, ClickHouse) have niche positions but have not scaled to threaten Snowflake's enterprise base. The competitive set is narrower in mid-2026 than it was two years ago.

Snowflake's stock-based compensation has been the loudest bear talking point for three years. SBC as a percentage of revenue ran above 40% at the IPO and has compressed to roughly 25% in the most recent fiscal year. The trajectory points to roughly 18-20% by FY27, which would bring the franchise into line with the broader hypergrowth software peer group.

The shift matters because the GAAP profitability path becomes visible at lower SBC ratios. Forward consensus models assume SBC compresses to roughly 18% by FY27, at which point the GAAP operating margin turns meaningfully positive. The bear case has assumed SBC stays elevated indefinitely. The trajectory does not support that.

This is the slowest-moving of the five points but it is the one with the most direct impact on the multiple. A franchise running at 30% revenue growth with normalising SBC and improving GAAP profitability supports a 90-100x forward multiple more comfortably than a franchise where the SBC trajectory is open-ended.

Sridhar Ramaswamy's transition from CEO of Neeva to Snowflake CEO in early 2024 created roughly six quarters of execution uncertainty. The strategic refresh that followed has now landed: AI workload prioritisation, Horizon Catalog repositioning, and a tighter capital allocation framework on M&A.

The execution has been clean. The product velocity has improved. The customer reception of the new strategic direction, judging by NPS and customer churn data, is positive. The CEO transition risk that was a meaningful overhang in 2024 is now off the table.

This is the kind of soft factor that does not show up in the consensus model but that institutional investors weight when sizing positions. The pattern in software franchises is that CEO transitions create a 12-18 month overhang on the multiple that compresses once execution is proven. Snowflake is past that overhang.