AI

Pentagon expands Scale AI contract to $500M, five-fold increase

Wednesday, May 6, 2026Read Original

Details

  • The U.S. Department of War's Chief Digital and Artificial Intelligence Office (CDAO) expanded its enterprise agreement with Scale AI from $100M to $500M on May 6, 2026, citing rapid uptake since the original award in September 2025.
  • Scale AI provides the Pentagon's data labeling, computer vision, and generative AI infrastructure; competitors in defense AI include Palantir Technologies ($10B+ in DoD ceiling), Anduril Industries ($20B Lattice AI C2), and xAI/OpenAI/Google/Anthropic (each $200M for agentic AI systems).
  • The CDAO Production Other Transaction Authority (OTA) model allows any DoD component to initiate Project Agreements without competitive re-solicitation, leveraging pre-negotiated pricing, CDAO co-funding, and streamlined contracting at classified network levels (NIPR, SIPR, JWICS).
  • Scale's platform spans the Data Engine (computer vision, EO/IR, SAR imagery labeling), GenAI Platform (fine-tuning on classified networks), Scale Donovan (decision-support for unstructured intelligence data), and capability sprints tailored to mission requirements.
  • Demand across Army, Navy, Marine Corps, and defense agencies exceeded the original $100M ceiling, signaling enterprise-wide AI adoption and validating streamlined procurement as a model for accelerating defense technology deployment.

Impact

This expansion accelerates the Pentagon's shift toward operational AI at the tactical edge, reducing deployment timelines from years to months and embedding vendor lock-in through centralized infrastructure decisions. The OTA model is now a template for defense AI procurement, likely driving similar expansions across Palantir, Anduril, and emerging contractors. Over 12–24 months, expect increased data labeling and fine-tuning demand, pressure on classified-network AI inference, and consolidation favoring platforms that integrate data, models, and decision-support into unified pipelines.

Rift Dispatchpractical systems & stories, weekly