US Army Exhausts 'Unlimited' Ask Sage AI Token Budget Weeks After Promise, Forcing New Limits

The US Army's Combat Capabilities Development Command (DEVCOM) informed employees in mid-June 2026 that the Army CIO's annual token budget for the Ask Sage generative AI platform had been exhausted, just weeks after promising "unlimited tokens" in May 2026, forcing the reimposition of usage limits for the Department of Defense-wide tool.

Rapid Adoption Overwhelms AI Resources

The swift depletion of the token budget highlights a significant challenge in scaling AI adoption within large government organizations. An anonymous Army employee noted that "the whole Army burned through the whole year of tokens for just one service," underscoring the high demand for generative AI capabilities. Despite the budget exhaustion, Army leadership characterized the rapid consumption as "robust and enthusiastic adoption" of AI tools.

Ask Sage Platform and Token Mechanics

Ask Sage is a multimodal generative AI platform utilized across the Department of Defense for various tasks, including personnel classification and acquisitions. The platform integrates models from leading AI developers such as OpenAI, Google, and Meta. Each year, the platform operates on an "enterprise pack" of 100,000,000 tokens. For context, approximately one token is equivalent to 3.7 characters.

Prior to the budget exhaustion, employees were allocated at least 200,000 tokens monthly, with an automatic top-up feature if their allotment was depleted. Furthermore, non-using employees reportedly received prompts encouraging them to increase their token consumption, contributing to the rapid expenditure.

Uncertainty Surrounds Future Token Allocation

An internal email circulated within the Army warned that the renewal of the Army CIO's token pool after October 1, 2026, remains uncertain. This lack of clarity raises questions about the long-term viability and funding models for large-scale AI deployments within the government. The incident also brings into focus the gap between the Pentagon's stated goals for AI adoption and the practical budgetary constraints involved in implementing such technologies.

Token Consumption in Military Operations

The Department of Defense has demonstrated a high capacity for AI token consumption in operational settings. During Operation Epic Fury in Iran, the DOD reportedly consumed 20 billion tokens per day over a 38-day period. This figure provides a benchmark for the potential scale of AI usage in military contexts and further emphasizes the need for robust and sustainable token allocation strategies.

Implications for Government AI Strategy

The rapid exhaustion of the US Army's AI token budget for Ask Sage underscores critical considerations for government agencies deploying generative AI. It challenges the sustainability of token-based pricing models when faced with widespread, enthusiastic adoption. The incident suggests that while the demand for advanced AI tools is high, the budgeting and resource allocation mechanisms may not yet be fully equipped to handle the scale of usage required for enterprise-wide deployment.

Conclusion

The US Army's experience with Ask Sage highlights the complexities of integrating advanced AI technologies into large organizations. The rapid depletion of an "unlimited" token budget within weeks of its announcement points to the need for more realistic forecasting and flexible budgeting for AI resources. As the Department of Defense continues its push for AI adoption, addressing these practical constraints will be crucial for ensuring the sustained and effective use of platforms like Ask Sage.

Sources

• 1 Introduction
• When AI Navigates the Fog of War
• Superhuman AI for Generals.io Using Self-Play Reinforcement Learning
• tokens.txt
• https://www.wired.com/story/the-army-is-burning-through-its-ai-tokens/