GOBLIN HOUSE
[ Enter Database → ]
Claim investigated: The Maven Smart System's integration across Pentagon infrastructure requires contract modifications and service orders that span multiple DoD component agencies, creating analytical blind spots for oversight researchers using standard USASpending queries Entity: US Department of Defense (Pentagon) Original confidence: inferential Result: STRENGTHENED → SECONDARY
The claim that Maven Smart System contracts span multiple DoD component agencies, creating analytical blind spots for USASpending queries, is highly plausible and consistent with established facts. The strongest case: Palantir's $10B enterprise agreement with the Army (W519TC25D0039) consolidates 75 separate contracts, and the Maven program itself shifted from a cross-functional team to an Army-led program of record. The Army also awarded Anduril a $20B consolidated agreement. These consolidations suggest that prior to consolidation, contracts were dispersed across sub-agencies (Army, SOCOM, etc.), each with its own FPDS codes, making aggregate searches under 'Department of Defense' incomplete. The strongest counter-case: USASpending does allow searching by parent agency (360000) or by contractor DUNS, which would capture all sub-agency awards to a given vendor. However, the claim's core insight—that standard queries fail—holds because most researchers use the generic 'Department of Defense' term without iterating over sub-agency FPDS codes. Underreported: The shift from dispersed to consolidated contracting is itself an opacity-creating event, as individual task orders and modifications become hidden within massive ceiling-value contracts.
Reasoning: The claim is strengthened by established fact #14: the Army consolidated 75 separate Palantir contracts into one enterprise agreement in July 2025. Prior to that consolidation, a USASpending search for 'Palantir AND Army' would yield scattered results, while a search for 'Palantir AND Department of Defense' would miss contracts awarded directly by SOCOM or other components. The Maven program's designation as an Army-led 'program of record' (fact #12) further supports the inference. The claim cannot reach primary confidence without direct evidence of a specific researcher's failed query, but the structural logic is consistent with known procurement patterns.
USASpending: Recipient: Palantir Technologies; Agency: Department of the Army (FPDS code 2100); Date range: 2020-2025; Exclude the consolidated contract W519TC25D0039
This would show the pre-consolidation dispersion of Palantir contracts across Army sub-components and confirm the analytical blind spot claim.
USASpending: Recipient: Palantir Technologies; Agency: Department of Defense (parent code 360000); Date range: 2020-2025; Filter by PIID containing 'MAVEN' or 'Maven'
This would test whether Maven-specific contracts are tagged with program keywords in USASpending, or are hidden within generic contract lines.
SEC EDGAR (Palantir 10-K filings): Palantir Technologies 10-K for FY2025; search text: 'Maven Smart System' or 'Project Maven' within 'Segment Revenue - Government'
Palantir's segment reporting may aggregate all DoD revenue into a single line, obscuring how much is specifically Maven-related—confirming the opacity claim.
Federal Procurement Data System (FPDS-NG) FOIA: Request records of all contract actions related to Maven Smart System (NAICS code 541512 or 541519) issued by Army, SOCOM, and other DoD components from 2020-2025
FOIA-requested FPDS data would reveal the exact distribution of Maven-related spending across sub-agencies, definitively proving or disproving the dispersion claim.
CRITICAL — The claim directly addresses a structural problem in how the largest military AI program in U.S. history (Maven Smart System, with $10B+ in planned spending) can be tracked for accountability. If the claim holds, it means Congress, GAO, and the public cannot determine the true cost or scope of Maven through standard public databases—an oversight failure that matters for budget authorization and for understanding the scale of AI-driven targeting systems.