Coverage from April through September shows analysts sharply divided on hyperscaler AI spending. Stanford economists estimate a nearly $1 trillion gap between hyperscaler AI spending and AI revenue collected since 2024. Goldman Sachs found AI revenues still fall short of what is needed to cover capex, and Reuters calculated capex could overtake free cash flow for Microsoft, Alphabet, Amazon, Meta and Oracle by 2027.
Bulls argue otherwise. Wells Fargo's Ohsung Kwon calls AI a "euphoric" bubble but says to ride it, Morningstar argued AI stocks traded at their largest discount since 2019, and Jefferies points to a roughly $2 trillion backlog. Total AI capex for 2027 is now estimated at over $1 trillion. Axios described the buildout as entering the show-me-the-money phase.
Source: Axios