ReportGem ReportGem 中文

GLOBAL RESEARCH ARCHIVE

High Yield & Loan Strategy: Funding the Future #1: Are we in an AI-capex bubble?

Published: 2026-06-25Institution: BofA Global ResearchPages: 30Original language: 英语Evidence page: 1

Research evidence excerpt

High Yield & Loan Strategy: Funding the Future #1: Are we in an AI-capex bubble?

Accessible version

High Yield & Loan Strategy

Funding the Future #1: Are we in an AI-

capex bubble?

This is the inaugural installment of our ‘Funding the Future’ series focused on credit market 25 June 2026

impact of the AI buildout. High Yield & Loan Strategy

United States

Not a bubble… yet

AI-led corporate capex is large, visible and accelerating, but sits only a touch above its Neha Khoda

Credit Strategist

70yr avg when scaled to GDP. Private investment, however, is elevated, but still leaves BofAS

room for an additional ~$400bn in annual spend to match historical peaks. Given the size +1neha.khoda@bofa.com646 855 9656

and scale of AI infrastructure demand, this threshold serves as our baseline, with an Adam Vogel

edge case reaching up to $750bn in annual incremental spending. Credit Strategist

BofAS

High base, thin cushion adam.vogel@bofa.com Dong Ba

The issue for credit is not the pace but the starting level of capex. Unlike past cycles Credit Strategist

that ramped from a low base, AI spend is building on elevated ground usually BofASdong.ba@bofa.com

characteristic of mid/late-cycle levels, on account of years of cheap money and capex- Elizabeth M. Han

friendly fiscal policy post GFC. That leaves less room for delayed monetization, cost Credit Strategist

overruns or a slower productivity payoff before investors start demanding a higher risk BofASelizabeth.han@bofa.com

premium. Yuri Seliger

The funding shock may be smaller than the spending shock BofASyuri.seliger@bofa.com

The strongest credit offset is that corporate America enters this cycle as a net saver,

with roughly 1.6% of GDP in excess capital that can be redeployed internally. That

The English excerpt is extracted automatically from the cited source page and may contain layout or recognition errors. It is never batch translated.

Open report viewer