AI and the US Economy: Accounting Investment and Production

Luisa Carpinelli, Filippo Natoli and Marco Taboga have published a working paper analyzing AI’s macroeconomic footprint in the United States and the role of data centers.

• Main announcement: The authors present an accounting framework and stylized AI production process showing formidable investment in data centers in 2025, and that IT and AI-related capital expenditure in the first three quarters of the year gave an outsized boost to aggregate demand while contributing less to GDP growth once the high import content of AI hardware is netted out.
• Details/background: The paper is a 35-page pre-print with 11 figures, available on arXiv (arXiv:2601.11196) with PDF/HTML/TeX source and an arXiv-issued DOI via DataCite; it estimates that services from new AI data centers could contribute to GDP over upcoming quarters at scales comparable to prior investment, and flags short reinvestment cycles and uncertainty about future AI demand as potential medium-term macro risks.