The bear case for AI capex no longer requires demand to disappoint.
The sharpest new bear vectors run on accounting clocks and market plumbing — Goldman's projection that depreciation fully offsets AI's earnings contribution by 2028, and GMO's scheduled equity-supply events implying a ~20% return drag — rather than on any failure of AI orders or revenue.
The same conclusion keeps arriving from across the workspace's research — 1 topics independently instantiate this theme. Filter the evidence by where it came from:
The sharpest new bear vector is pure market plumbing — credit spreads and deleveraging risk repricing the complex with zero failure of AI orders or revenue.
The sharpest bear vector is now pure break-even arithmetic demanding 4–10x revenue growth, advanced even while every observed demand signal prints hot.
GMO's dated equity-supply events — SpaceX lockups and lab IPOs adding roughly 5% to market cap — are a break mechanism requiring no failure of AI demand.
Burry's sharpened bear runs on financing structure, rates, and depreciation accounting rather than any demand failure, while AWS hikes and Anthropic's accelerating run-rate keep the demand leg printing.
The reversal mechanism is now depreciation arithmetic and earnings quality on a dated schedule, independent of whether demand holds.