OpenAI-Pentagon $200M OTA Contract for Classified Military AI

The Contract

On February 27, 2026, the Department of Defense executed an Other Transaction Authority (OTA) agreement with OpenAI with a $200 million contract ceiling for the deployment of AI systems in classified military environments. The contract covers “artificial intelligence capabilities for intelligence analysis, operational planning, and decision support across classified networks and environments.” The agreement was signed the same day the Pentagon issued its supply chain risk designation against Anthropic and the same day President Trump ordered all federal agencies to cease using Anthropic technology.

OTA agreements are a procurement mechanism that allows the DoD to bypass the Federal Acquisition Regulation (FAR) and its competitive bidding requirements. Originally designed to engage nontraditional defense contractors and accelerate technology adoption, OTAs have faced increasing Congressional scrutiny for their use as a vehicle to avoid transparency and competition requirements. The OpenAI contract was awarded without competitive bidding.

Contract Structure

The agreement is structured as a base contract with multiple option periods and task orders, allowing rapid expansion of scope without additional competitive procurement. Bloomberg reported that the contract has been amended at least three times since its initial execution, though the specific amendments and their content have not been publicly disclosed. The OTA structure means the contract is not subject to the same public disclosure requirements as FAR-based contracts, limiting Congressional and public visibility into its terms, deliverables, and performance.

OpenAI’s role under the contract includes deploying models on classified networks — an operational capability that requires facilities clearances, personnel clearances, and air-gapped infrastructure that OpenAI did not previously maintain. The contract reportedly includes provisions for OpenAI to establish a classified computing environment within existing DoD infrastructure, with OpenAI personnel operating under DoD security protocols.

The Timing Question

The simultaneous execution of three actions on February 27 — Anthropic’s designation, Trump’s cease order, and the OpenAI contract signing — has drawn scrutiny from Congressional oversight committees and legal analysts. The coordination suggests pre-planning across multiple government offices: the supply chain designation required CIO-level analysis, the executive directive required White House coordination, and the OTA required contracting officer execution. Senator Mark Warner called the timing “either a remarkable coincidence or evidence of a coordinated campaign to replace one AI provider with another for reasons that have nothing to do with national security and everything to do with political loyalty.”

Emil Michael’s Role

Reporting by The Intercept and Bloomberg identified Emil Michael — former Uber SVP and OpenAI board observer — as a key intermediary between OpenAI leadership and Pentagon decision-makers in the months preceding the contract. Michael’s connections to both Silicon Valley executive circles and defense establishment figures positioned him as a broker for the relationship. His involvement has raised questions about whether the contract reflects a genuine competitive assessment of AI capabilities or a relationship-driven procurement decision.