Large enterprise AI adoption declined 13% since July 2025 peak (US Census data) (www.narev.ai)

U.S. Census data from the Business Trends and Outlook Survey (BTOS) shows AI use among large enterprises (250+ employees) peaked in July 2025 at 13.4% then fell to 11.7% by early September—a 13% relative decline—breaking a steady growth trend that began in late 2023. The BTOS tracks responses from about 1.2 million U.S. businesses; this analysis uses data from Nov. 19, 2023–Sept. 10, 2025 and a rolling 6‑survey (12‑week) average to smooth short-term volatility. All firm-size categories fell since July except the smallest firms (1–4 employees), which rose from 9.65% to 10.3%. The survey asks whether a business used AI in producing goods or services in the prior two weeks (examples: machine learning, NLP, virtual agents, voice recognition). The dip is significant because it’s occurring among the best-resourced organizations, suggesting a pullback in enterprise deployments rather than broader market stagnation. Possible explanations include integration headaches, cost pressures, governance/regulatory caution, project failures, or a shift from experimentation to selective production use—each with different implications for vendors, talent demand, and investment. Methodological caveats matter: these are self‑reported short‑term use snapshots and modest absolute percentages (peak ~13%) that nonetheless signal a change in momentum worth watching as firms reassess AI strategy and ROI.