Demand forecast model for artificial intelligence

(a) Development of model and methodology The Secretary of Defense shall, acting through the Chief Digital and Artificial Intelligence Officer, develop a unified, Department-wide demand forecast model and methodology for consolidating and assessing artificial intelligence demand data across the Department of Defense, including data from Department-wide generative artificial intelligence platforms and applications and artificial intelligence capabilities fielded or operated by the military departments and defense agencies. (b) Required actions In carrying out subsection (a), the Chief Digital and Artificial Intelligence Officer shall— (1) establish an automated process for collecting and aggregating artificial intelligence program and budget data from the military departments and defense agencies, using the artificial intelligence activities data elements and guidance issued in response to section 1533 of the Servicemember Quality of Life Improvement and National Defense Authorization Act for Fiscal Year 2025 (Public Law 118–159; 10 U.S.C. 221 note); (2) develop a demand-based forecasting model for artificial intelligence requirements that incorporates metrics, including token consumption, compute utilization, throughput, response latency, user and transaction volume, and such other metrics as the Secretary determines appropriate to assess artificial intelligence demand; (3) update applicable policy guidance and instructions for use by all elements of the Department elements and military departments on the requirements for the demand forecasting, associated data collection, and analysis; and (4) provide the Director of Cost Assessment and Program Evaluation direct access to all artificial intelligence demand data, the forecasting model required under paragraph (2), and all underlying raw data and supporting metrics used to develop and update that model, at a cadence sufficient to support annual budget submissions and future years defense program development. (c) Briefings (1) Initial briefing Not later than September 15, 2027, the Secretary, acting through the Chief Digital and Artificial Intelligence Officer, and the Director of Cost Assessment and Program Evaluation shall jointly provide to the congressional defense committees a briefing on— (A) the methodology used to consolidate artificial intelligence demand data and the metrics used to quantify that demand; (B) the policies and processes established to provide the Director of Cost Assessment and Program Evaluation direct access to artificial intelligence demand data and the demand-based forecasting model pursuant to subsection (b)(3); and (C) the status of the demand-based forecasting model, including any gaps in data availability or quality affecting budget forecasting for artificial intelligence requirements across the future years defense program. (2) Annual briefings Concurrent with the delivery of budget materials to Congress under section 1105 of title 31, United States Code, for each of fiscal years 2028 through 2031, the Secretary, acting through the Chief Digital and Artificial Intelligence Officer, shall provide to the congressional defense committees a briefing on current artificial intelligence demand data across the Department and forecasted artificial intelligence demand for each fiscal year of the then-current future years defense program. (d) Definitions In this section: (1) The term future years defense program means the program a program submitted under section 221(a) of title 10, United States Code. (2) The term token means the basic unit of text or data processed by a generative artificial intelligence model, used as the basis for measuring computational usage and associated costs.