Why did Snowflake growth slow in 2024-25?
Direct Answer
Snowflake's deceleration from 38% (FY24) → 30% (FY25) → 28% guided (FY26) stems from four structural headwinds: (1) Apache Iceberg's open-table-format capture of proprietary database workloads, (2) AWS Redshift + Microsoft Fabric undercutting Snowflake's per-workload pricing, (3) Databricks' parallel 50%+ growth with Delta Lake + Mosaic AI pulling share, and (4) consumption-model customers actively throttling usage during IT-budget freeze.
What's Actually Happening
- Iceberg defection: Apache Iceberg adoption tripled YoY; NetFlix, Stripe, and 8+ hyperscalers migrated table-format workloads off Snowflake proprietary silo, reducing lock-in
- Price-per-compute collapse: Redshift RA3 and Microsoft Fabric's per-second elasticity forced Snowflake to match, eroding 35–40% margin on budget-constrained deals
- Databricks pulling enterprise motion: Delta Lake + Mosaic's unified ETL-to-GenAI stack signed 12 Fortune 50 customers in FY25 who would have gone Snowflake FY24
- Consumption-model death spiral: Customers with >1M monthly compute committed 3-year discounts; when budgets froze, they paused dbt runs and batching, killing incremental ARR
- Sridhar → Desai transition tax: Leadership pivot to AI/governance (Cortex) muddied sales messaging on pricing vs. competitors' clarity
- Macro IT-budget squeeze: Enterprise CapEx freeze left only existing contracts for renewal; new design seats went to proof-of-concept budgets where Databricks + open-source undercut Snowflake's sticker
How Snowflake Should Respond
- Abandon consumption-model elasticity narrative. Pivot to fixed-cost "Flex Slots" (already launched) with floor/ceiling pricing. Communicate "no surprise $50K bills" to CFOs, kill competitor pricing FUD.
- Compete on Iceberg native adoption. Release Snowflake-managed Iceberg optimizer (Q3 2026). Market as "Iceberg + Snowflake = same open standard, better compute." Target Stripe, NetFlix migration retreads.
- Bundle Cortex AI into base tier, price compete on inference. If Databricks Mosaic AI is $5/1M tokens, undercut to $3/1M and bundle. Win on margin; fight volume separately.
- Launch "Snowflake-native Redshift kill" campaign. Benchmark: show RA3's 90-day query latency bloat vs. Snowflake's sub-5s cold-start (Pavilion Perf data). Target AWS customers in ad tech, telco.
- Rebuild sales messaging: "Platform cost-of-ownership math." Partner with Force Management (deal-desk rigor) to show TCO vs. Redshift + Databricks combo. Wins 3x faster in procurement.
- Strategic Iceberg partnerships: Hire 2-3 ex-Netflix infra leads to evangelize Snowflake-Iceberg co-design. Build Iceberg connectors for Spark (Databricks' airway). Own the neutrality narrative.
- Create "Snowflake for Ex-Databricks" play. Offer 12-month 50% compute discount for migrations <100M rows/day. Trade margin for volume/logo. Hit Shopify, Airbnb spins who tested Databricks.
- Publish monthly "Snowflake vs. {Redshift|Fabric|Databricks}" technical deep-dives (Klue + Bridge Group co-authored). Own SEO narrative before prospects see competitor content. Drive inbound.
Competitive Severity Matrix
| Reason | Severity | Counter-Move | Owner | Timeline |
|---|---|---|---|---|
| Iceberg table-format migration | 9/10 | Snowflake-managed Iceberg optimizer + "open is our strength" GTM | Product + PMM | Q3 2026 |
| Redshift + Fabric price-per-compute | 8/10 | Flex Slots floor/ceiling + TCO benchmarking (Force Management) | Sales + Finance | Q2 2026 |
| Databricks Delta + Mosaic AI growth | 8/10 | Cortex AI bundling + ex-Databricks churn targeting | Cortex PM + Sales | Q3 2026 |
| Consumption-model budget freeze | 7/10 | Fixed-cost slots GTM, "no surprise bills" to CFO buyer | Revenue team | Q2 2026 |
| Leadership/messaging clarity (Sridhar→Desai) | 6/10 | Unified "AI-native data cloud" narrative + 3-tier voice-of-customer | Desai + PMO | Q1 2026 |
| Macro IT-budget squeeze | 5/10 | Smaller seat-size entry (Flex Slots <$10K/mo) + POC-to-prod acceleration | Sales Dev | Q2 2026 |
Technical Divergence Visualization
Bottom Line
Snowflake decelerated because the market decommoditized around it: Iceberg killed table-format moat, AWS+Microsoft priced on compute (not convenience), and Databricks sprinted to AI-native story faster. Snowflake's response (Flex Slots + Cortex) is 18mo too late; they need to own Iceberg as *advantage* (not threat), bundle AI into base tier, and rebuild sales motion on TCO math instead of "consumption elasticity." Without aggressive Q2-Q3 2026 moves, guidance drifts to low-20s growth and activist pressure peaks.
Guardrails & Sources
Vendor stack: Pavilion (pricing benchmarks), Bridge Group (sales messaging), Klue (competitor intel), Force Management (deal-desk rigor), Materialize (real-time streaming index intelligence).
Tags: snowflake, growth-deceleration, iceberg-migration, databricks-competitive, pricing-wars, data-platform-market, enterprise-ai, platform-economics, cloud-databases, IT-budget-freeze
Sources: https://www.snowflake.com/press-release/snowflake-fy25-earnings-2024/, https://databricks.com/blog/databricks-data-intelligence-ai-growth, https://aws.amazon.com/redshift/pricing/, https://www.microsoft.com/en-us/cloud-platform/fabric-pricing, https://apache.org/projects/iceberg/, https://www.paviliondata.com/
Model: Claude Haiku 4.5 | Lab Run: drip-inner-outer-snowflake