Net Dollar Retention Rate Decomposition: What Drives 139% NRR?

AIP is bundled with Foundry and Gotham at no separate charge — it is not sold as a distinct module with its own SKU. This means NRR expansion is driven by usage growth (more users, more use cases, more compute consumption) rather than by unlocking a new product line. The pricing model is multi-dimensional: contracts are sized by user count, deployment scope, data volume, and compute consumption. Enterprise customers sign multi-year, fixed-fee agreements covering the initial deployment scope, with expansion renegotiated as the footprint grows.

The bootcamp-to-production funnel compresses the traditional enterprise sales cycle from twelve months to days. Approximately 70-75% of bootcamp participants convert to paid contracts within one quarter, with over 1,300 bootcamps completed by June 2024. Post-bootcamp deal sizes are substantial: a healthcare company signed $96M TCV after two summer bootcamps; an engineering services company signed $80M after fall demos. These large initial contracts then expand as customers add use cases.

The expansion follows a consistent pattern: organizations start with a single department and a handful of use cases, then expand user access and workflow coverage across the enterprise. Lear expanded from 100 users and 4 use cases to 16,000 users and 280 use cases. A utility company grew ACV from $7M to $31M within a single year. A medical device manufacturer grew ACV more than eightfold within five months. Management attributed expansions to 'value generated from new use cases' — confirming that the expansion engine is use-case proliferation, not price increases.

Palantir does not disclose segment-level NRR. The reported 139% is a blended figure. However, segment revenue growth differentials allow reasonable inference: US Commercial grew 137% YoY while US Government grew 66% YoY in Q4 2025 — a 71-percentage-point gap. Since NRR measures year-over-year revenue from the same customer cohort, commercial accounts are expanding at roughly double the rate of government accounts within the existing base.

Government NRR is structurally constrained by fixed-price, multi-year contract architecture. Management disclosed that RPO calculations exclude 'contracts with an initial term of less than 12 months and contractual obligations that fall beyond termination for convenience clauses, both of which are common in most of our government business.' Government expansion comes through contract ceiling increases (Maven raised from $480M to $1.3B on 'surging demand') and new agency rollouts, not organic usage-based growth.

Maven's Program of Record designation in March 2026 changes the government revenue risk profile: it converts discretionary pilot spending to baselined multi-year defense funding. The Army's $10B enterprise agreement further transitions government revenue from fragmented annual procurement to durable institutional spending. Neither contract guarantees ceiling utilization — both are IDIQ structures — but POR status substantially de-risks the government NRR floor.

International commercial is the weakest NRR contributor, growing only 8% YoY in Q4 2025 against 137% US commercial growth. Sovereign procurement constraints, data localization requirements, and Palantir's deliberate decision to concentrate sales bandwidth in the US market explain the gap. CEO Karp acknowledged 'we don't have bandwidth outside America.' The UK government's March 2026 signal toward domestic tech alternatives illustrates the sovereign preference risk that limits international government NRR expansion.

Enterprise SaaS NRR compression is a near-universal pattern: every major platform that achieved 130%+ NRR has subsequently compressed toward 115-125% as the customer base matured. Snowflake compressed from 174% to 125% over 13 consecutive quarters of decline (Q1 FY2023 to Q3 FY2026), driven by customer base maturation, consumption optimization, and the law of large numbers. Twilio compressed from 126% to 102% over 8 quarters. CrowdStrike compressed from 125%+ to 112% before recovering to 115%.

The bull case: Palantir's use-case proliferation model (Lear went from 4 to 280 use cases) creates continuously expanding TAM within each customer, unlike consumption-based models where workloads eventually optimize. The bear case: Optifai data from 939 B2B SaaS companies shows enterprise median NRR of 118%, with top quartile at 130% — suggesting 139% is 2+ standard deviations above the sustainable level.