Big Tech keeps $300B of AI exposure off balance sheets

by@FT

Summary

Big Tech companies are utilizing special accounting practices to keep approximately $300 billion of their AI exposure off their balance sheets, leveraging special purpose vehicles and guarantees under current regulations. This maneuver has raised concerns as chipmakers like Nvidia and Broadcom increasingly bear the risk as ultimate guarantors for AI-lease financing structures, thus concentrating substantial downside risk in a few suppliers without formally acknowledging these liabilities. Regulatory authorities are now scrutinizing these off-balance-sheet commitments, drawing parallels to historical corporate failures and assessing whether stricter disclosure and recognition rules are necessary for such guarantees.

Analysis

Big Tech: “Big Tech” in this context refers to large US-based technology and cloud infrastructure companies such as Alphabet, Amazon, Microsoft, Meta, Oracle, Nvidia, Broadcom, and other major AI and data-center players that dominate global computing and AI infrastructure. These firms are collectively structuring AI-related data center leases, chip purchase obligations, and residual value guarantees in ways that keep a large portion of their AI financing exposure classified as off-balance-sheet contingent commitments rather than conventional debt. Risk_concentration: Analyses over the past month highlight that chipmakers like Nvidia and Broadcom increasingly act as ultimate guarantors for AI-lease financing structures, concentrating substantial downside risk in a handful of suppliers even though many of these guarantees are not recorded as liabilities. Accounting_structure: Recent reporting describes how hyperscalers use special purpose vehicles, uncommenced leases, and residual value guarantees so that large AI infrastructure commitments remain contingent and off their main balance sheets under current accounting rules. Regulatory_attention: Commentary from accounting and finance experts in recent articles notes that standard setters and regulators are scrutinizing these off-balance-sheet AI commitments, comparing some structures to devices associated with past corporate failures and debating whether disclosure and recognition rules should be tightened for guarantees and variable interest entities.

Categories

aitech

Related sources

View Original Tweet