How Is the AI Build-Out Financed? Internal Funds, Leases, and Depreciation Compared with the Dot-Com Boom
Preprint
- 20 October 2026
- preprint
- Published by Elsevier in SSRN Electronic Journal
Abstract
We compare the financing of the AI infrastructure build-out with that of the dot-com build-out, using one selection rule and one cash measure: capital expenditure minus operating cash flow, scaled by sales (the gap). Through 2025 the ten largest capital spenders in the build-out industries generated more operating cash than they spent on capital, in aggregate; their dot-com counterparts did not by 2001. The aggregate hides dispersion. In the four quarters to mid-2026 Oracle’s capital spending was 1.74 times its operating cash flow and Amazon’s slightly exceeded it. The five hyperscalers borrowed $234 billion net and added $224 billion to cash. Their $1.09 trillion of leases not yet begun would raise their gap from -6.8% to about -3% of sales once payments start; treated as purchases spread over three to ten years, the gap would be between -3.6% and +6.8%. Faster server depreciation would lower 2025 operating income by a few percent. In daily data for 1996–2025, the gap does not reliably condition tail co-movement between the builders and the rest of the market.Keywords
This publication has 24 references indexed in Scilit:
- Bubbles for FamaJournal of Financial Economics, 2019
- Leveraged bubblesJournal of Monetary Economics, 2015
- On the network topology of variance decompositions: Measuring the connectedness of financial firmsJournal of Econometrics, 2014
- Bubbles, Financial Crises, and Systemic Risk2013
- Understanding earnings quality: A review of the proxies, their determinants and their consequencesJournal of Accounting and Economics, 2010
- Was there a Nasdaq bubble in the late 1990s?☆Journal of Financial Economics, 2006
- Overconfidence and Speculative BubblesJournal of Political Economy, 2003
- Testing the pecking order theory of capital structureJournal of Financial Economics, 2003
- Depreciation-policy changes: tax, earnings management, and investment opportunity incentivesJournal of Accounting and Economics, 1999
- A comparison of the Two One-Sided Tests Procedure and the Power Approach for assessing the equivalence of average bioavailabilityJournal of Pharmacokinetics and Biopharmaceutics, 1987