The U.S. Army has hit a limit on its generative AI usage after exhausting its annual pool of tokens by mid-June 2026, just weeks after the Department of Defense claimed widespread AI adoption across its workforce. Employees at the Army's Combat Capabilities Development Command received an internal email stating that unlimited token access, announced in May, had been suspended due to depletion of the Army CIO's allocation. The AI platform in use, Ask Sage, supports multiple large language models including Gemini, Llama, and ChatGPT, and is used for tasks like reclassifying personnel descriptions. Workers were initially given at least 200,000 tokens per month, with automatic top-ups, and were encouraged to increase usage—even those not actively using their allocation received prompts to do so.
Ask Sage operates on a token system where one token equals about 3.7 characters, and the Army's total annual access was 100 million tokens. Despite this, demand surged, leading to renewed usage caps and uncertainty about renewal beyond October 1, 2026. The Army and DOD did not respond to requests for comment, and it remains unclear whether tokens for classified operations draw from the same pool. This surge in consumption mirrors broader trends, as other organizations like Meta and Uber have also moved to restrict internal AI token use after initial permissiveness. The Pentagon continues to prioritize AI integration, recently reducing staff at the Civilian Protection Center of Excellence while developing an AI tool to replace some of its functions.
The Army's rapid depletion of AI tokens suggests a gap between enthusiasm for new technology and realistic resource planning. Pushing widespread AI use without scalable infrastructure risks creating dependency on tools that may not be sustainably funded or managed. If token limits return, frontline AI adoption could stall despite official encouragement.
Editorial note: AI-assisted opinion, not established fact. Full disclaimer →
