Snowflake Inc.

Snowflake sells the AI Data Cloud — a cloud data platform for analytics, data engineering, data science, collaboration, application development, and secure data sharing. Management describes it as a single source of truth that lets customers consolidate data without copying or moving the underlying data.

Snowflake sells the AI Data Cloud — a consumption-based data platform that consolidates enterprise data into a single source of truth for analytics, AI, and data sharing. The latest 10-Q says Snowflake also supports external customer-controlled environments via open data formats like Apache Iceberg.

Snowflake sells the AI Data Cloud — a governed enterprise data platform for analytics, AI, secure sharing, and application workloads. The business is still fundamentally a control plane: policy, lineage, access controls, and workflow state sit above the data.

• monetization model: Consumption-based cloud/SaaS; product revenue from usage credits + professional services. (10-K FY2025, acc. 0001640147-25-000032, Business/Revenue; sec.gov )
• revenue mix: Product revenue ~95% of total; professional services/other ~5% (FY2025); one reportable segment. (same 10-K, segment/revenue note)
• who pays: Enterprise organizations; contracting entity pays for consumption. Internal budget owner not disclosed. (same 10-K, Business/Customers)
• margin_value_capture: GAAP product gross margin ~75%, total gross margin ~69%; GAAP operating margin negative; value accrues in product gross profit before S&M/R&D. (same 10-K, MD&A)

• customer classes: Platform & Developer Teams, data/analytics executives (CDO/CIO/CTO), line-of-business analysts, CISOs & Security Operations for security/governance, Compliance & GRC Officers for regulated workloads. (10-K FY2025, acc. 0001640147-25-000032, Customers; sec.gov )

Snowflake’s monetization mechanics are usage-based:

• Compute is consumed in credits based on virtual warehouse size and runtime.
• Storage is billed separately, typically per terabyte per month.
• Data transfer is billed based on movement across regions or cloud providers.
• Customers can buy on-demand or commit to capacity contracts, which produce remaining performance obligations, or RPO.

This is meaningfully different from a fixed-seat SaaS model. A customer that increases queries, data volume, data sharing, or workloads increases spend without repurchasing seats. But it also means Snowflake is exposed to customer optimization efforts: if a customer rightsizes compute, pauses warehouses, or reduces idle query load, recognized product revenue can decelerate or temporarily decline.

FY2026 revenue was $4.684B (+29.2% YoY), with gross profit $3.146B (67.2% gross margin) and operating loss -$1.435B (-30.6% operating margin). Q1 FY2027 revenue was $1.391B (+33.5% YoY), product revenue was $1.33B (+34% YoY), and the company reported $9.21B of remaining performance obligations and 779 customers with trailing 12-month product revenue above $1M.

Q1 FY2027 revenue was $1.391B (+33% YoY), product revenue was $1.33B (+34% YoY), gross profit was $926M (66.7% gross margin), and operating loss was $326M (-23.4% operating margin). Snowflake ended the quarter with 13,912 total customers, 779 customers with >$1M trailing-12-month product revenue, 813 Forbes Global 2000 customers, 126% NRR, and $9.21B of remaining performance obligations.

FY2026 revenue was $4.684B, gross profit was $3.146B (67.2% gross margin), and operating loss was -$1.435B (-30.6% operating margin). Q1 FY2027 revenue was $1.391B, gross profit was $926M (66.7% gross margin), and operating loss was -$326M (-23.4% operating margin).

Snowflake monetizes the control plane of enterprise data: governance, lineage, permissions, clean rooms, and AI context. The more deeply a customer encodes policy, semantics, and AI workflows inside Snowflake, the more expensive it becomes to move to another stack.

The moat is the policy-and-context layer, not the raw data bytes. Snowflake captures value where enterprises encode governance, lineage, access control, clean rooms, and AI workflows. The open-format strategy broadens reach, but it also means the moat depends on workflow gravity and revalidation cost rather than data custody alone.

Snowflake monetizes the friction of enterprise replatforming. When customers encode permissions, lineage, clean rooms, semantic context, and AI workflows inside Snowflake, the migration burden is not storage copy cost — it is revalidation, retraining, and parallel-run risk.

• 10-K says the platform enables customers to consolidate data into a "single source of truth" and share it securely "without copying or moving the underlying data"; the architecture uses a single persistent copy plus data clean rooms. Source: FY2026 10-K.
• FY2026 companyfacts from SEC show revenue of $4.684B, gross profit of $3.146B, and operating loss of -$1.435B. Source: SEC XBRL companyfacts / FY2026 10-K.
• Q1 FY2027 EX-99.1 earnings release reported revenue of $1.39B, product revenue of $1.33B, NRR of 126%, RPO of $9.21B, 779 customers with >$1M TTM product revenue, 813 Forbes Global 2000 customers, and 616 net new customers. Source: FY2027 Q1 EX-99.1.
• The same release said more than 13,600 accounts are now using Snowflake AI capabilities and 7,100+ accounts use Cortex Code; the company also raised FY2027 product revenue guidance to $5.84B. Source: FY2027 Q1 EX-99.1.
• 10-K risk factors explicitly note that support for open data formats like Apache Iceberg can reduce switching costs between Snowflake and competitors. That is the main counterweight to the lock-in thesis. Source: FY2026 10-K.

• SEC-disclosed large-customer base: 779 customers with >$1M TTM product revenue.
• SEC-disclosed RPO: $9.21B.
• RPO per >$1M customer equivalent: about $11.8M ($9.21B / 779).
• Conservative migration model for a representative >$1M annual-spend customer: 12 months of overlap plus revalidation/retraining implies a switching burden of roughly $2-4M, or ~2-4x the annual spend floor. This is an estimate, not a reported figure, but it is anchored to the SEC-disclosed spend floor and the platform's single-copy / governed architecture.

• The 10-Q says Snowflake enables customers to consolidate data into a single source of truth and that 97% of Q1 revenue came from existing customers under capacity arrangements.
• The Q1 FY2027 release reported 126% NRR and 779 customers with >$1M trailing-12-month product revenue; those customers represented about 68% of product revenue.
• RPO was $9.21B, with roughly 50% expected to be recognized in the next 12 months; weighted-average remaining life of capacity contracts was 2.6 years.
• The 10-Q explicitly warns that open data formats like Apache Iceberg let customers process data in external customer-controlled environments outside Snowflake, reducing lock-in.

• RPO / quarterly product revenue = $9.21B / $1.33B ≈ 6.9x.
• RPO / large-customer count = $9.21B / 779 ≈ $11.8M per >$1M customer equivalent.
• If 50% of RPO is recognized in the next 12 months, then roughly $4.6B of contracted revenue is already spoken for in the next year, and the weighted-average remaining life of capacity contracts (2.6 years) sets a multi-year unwind window.
• Practical switching burden: a representative $1M+ annual-spend enterprise must run overlap + policy revalidation + workflow migration, making a sub-12-month escape unlikely without major business disruption.

• The 10-Q says 779 customers with trailing-12-month product revenue > $1M represented approximately 68% of product revenue. That concentration means the highest-value accounts are the core monetization engine.
• The same 10-Q says net revenue retention was 126% as of Apr. 30, 2026; 13,912 total customers were active; remaining performance obligations were approximately $9.2B, and roughly 50% of that is expected to be recognized in the next 12 months.
• The 10-Q also says the weighted-average remaining life of capacity contracts was 2.6 years, which makes churn an unwind process, not a one-quarter event.
• The Q1 FY2027 earnings release said revenue grew 33% YoY, 13,600+ accounts were using Snowflake AI capabilities, Cortex Code was in use across 7,100+ accounts, and the company added 616 net new customers.

The 10-Q implies a large-account spend pool of at least $779M annually (779 customers x $1M minimum each), and those accounts are actually the majority of product revenue (~68%). A representative enterprise migration therefore has to displace at least $1M of billed annual spend and then pay for overlap, revalidation, and retraining. Even a conservative 6-12 month dual-run implies roughly $0.5M-$1.0M of duplicate platform spend before labor/compliance costs; for heavier regulated workloads the unwind cost is materially higher. The 2.6-year average contract life makes this a multi-year switching wall rather than a one-quarter churn event.

`

Average product revenue per customer

= $3,528,000,000 / 11,618

= $303,676 per customer

Next-year product revenue from current customers

= $303,676 × 1.26

= $382,632 per customer

Next-year gross profit from current customers

= $382,632 × 0.767

= $293,478 per customer

`

This calculation shows that, before allocation of S&M, R&D, and G&A, an average existing Snowflake customer generates roughly $293K of gross profit in the following year under the assumption that all current customers expand at the overall NRR rate and no churn occurs.

That is not an LTV/CAC calculation. It is a high-level expansion-unit calculation. It ignores new customers, churn, customer mix by size, industry margin differences, and the fact that NRR is an average that may be higher for large enterprise customers and lower for smaller customers.

Product gross profit per average existing customer:

The main non-GAAP gross-margin sensitivity is less dramatic than the NRR sensitivity. Snowflake’s unit economics are very sensitive to net revenue retention, which has been falling from triple-digit early-stage expansion toward the mid-120s.