Major cloud and data center operators are committed to a projected billions–billions AI capital expenditure cycle in 2026.1 If AI workload monetization fails to match that spending pace, hyperscalers face write-downs on stranded assets, balance sheet deterioration, and compressed valuations.
The buildout spans data center construction, AI-optimized compute hardware, and supporting infrastructure — including power management components sourced from chipmakers. These contracts are largely irreversible once signed. Operators cannot easily scale back mid-cycle.
The core earnings risk is structural. AI-optimized hardware and data centers carry multi-year depreciation schedules. If the assets sit underutilized, depreciation hits earnings without offsetting AI revenue. That is the stranded asset scenario: capital deployed, returns delayed or absent, losses accelerated.
Write-downs compress earnings per share immediately. Hyperscaler stocks trade at premium multiples anchored to AI growth expectations.1 A write-down announcement delivers a double blow — lower reported earnings and downward pressure on the earnings multiple investors are willing to pay. Valuation math turns negative fast.
Investor backlash is a separate amplifier. Markets have already priced significant AI upside into hyperscaler equities. Any signal that capex is running ahead of monetization erodes that premium. Capital rotation out of overextended AI infrastructure names becomes likely.
The risk assessment classifies severity as catastrophic and likelihood as medium.1 Medium likelihood means this is not a tail risk scenario — it is a plausible base case if enterprise AI adoption ramps more slowly than the capital cycle demands.
Three leading indicators matter for investors tracking this risk. First, quarterly AI revenue disclosures — operators that do not separately break out AI workload revenue make monetization progress opaque. Second, capex guidance revisions — any reduction in forward spending signals internal concern about near-term returns. Third, data center utilization rates — low utilization on AI-optimized infrastructure is an early warning before write-downs appear on balance sheets.
The asymmetry is the central problem for equity positioning. AI monetization upside is largely priced into current valuations. Write-down and multiple compression downside is not.
Sources:
1 Via News AI Infrastructure Risk Assessment, May 20, 2026
