Preprint2026

Risk Repricing and Realized Capital Spending: Evidence from AI Infrastructure Firms

Fuli Yang; Jianping Huang; Jing Huang

Energy & Compute Infrastructure →

Abstract

When artificial intelligence infrastructure firms come under market pressure, how much can prices tell us about their subsequent spending? Oracle shares lost 10.51 percentage points relative to the Invesco fund tracking the Nasdaq-100 in the first regular session after its December 2025 earnings release. Company-wide cash capital expenditure rose 54.9% in the fiscal quarter already under way, fell 11.5% in the next quarter and then rose 72.8%. In autumn 2026, Microsoft, Meta and Oracle paid 45, 110 and 275 basis points over Treasuries on new five-year debt. Their previously reported operating cash flows covered 1.44, 1.27 and 0.62 times trailing cash investment. In a separate, earlier window, daily observations on outstanding bonds show isolated stress crossings around earnings. A joint equity–credit tail measure adds little to five-day spread forecasts in that sample. These observations document a selloff alongside continuing investment and substantial differences in borrowing terms. They leave open which market, if any, reliably anticipates a spending turn.

Keywords

JEL: G12, G14, G31

Cite this paper

Yang, F.; Huang, J.; Huang, J. (2026). Risk Repricing and Realized Capital Spending: Evidence from AI Infrastructure Firms. GARCH Press. https://doi.org/10.69057/WP-0135-V1

Version & publication history

  1. Original repository record

    Public catalogue record. The manuscript is available to authorized readers.

  2. Added to GARCH Research Papers

    Bibliographic record added. This is not a new manuscript release or a new DOI.

Access & rights

Sign in to preview or download the PDF. Access is available to authorized accounts.