OPEN
Despite its expensive price tag, the technology is nowhere near where it needs to be in order to be useful. Overbuilding things the world doesn't have use for, or is not ready for, typically ends badly.
Source: Institutional Investor ·
Analyst note
Covello’s argument is fundamentally microeconomic: if marginal dollars buy capability that does not clear adoption hurdles, the capital cycle can break before science stops improving. Bulls rejoin that infrastructure eras look overbuilt until applications arrive.
We do not treat this as a dated mechanical forecast of a ‘burst year’—markets rarely cooperate with nouns. The operational question is whether incremental AI revenue can cover incremental AI compute at scale without permanent strategic subsidy. Watch vendor gross margins, hyperscaler cloud growth composition, and enterprise renewal commentary.
Evidence timeline
Goldman’s senior U.S. equity researcher publicly advanced skepticism that AI spend would quickly earn its cost of capital, emphasizing mismatch between price and usefulness.
Equity markets showed episodes of concentrated AI-adjacent multiple compression, but index-level drawdowns were not uniquely AI-attributable absent rate shocks.
Mid-2026 outcomes remained inconsistent with a clean ‘bubble resolved’ verdict: revenue was real in segments, while cash-flow sustainability was still contested.