The pattern reveals a structural tension in AI infrastructure investing. Capital markets have priced AI chip demand at maximum growth velocity. Any guidance landing in-line triggers outsized selling. The gap between "good" and "good enough" now generates double-digit drawdowns.
This is not a demand story. AI infrastructure spending continues. Hyperscalers are still building data centers and chip orders remain robust. The issue is valuation: semiconductor stocks have absorbed so much future growth that the margin for disappointment has collapsed.
Broadcom's two-day collapse illustrates the asymmetry directly.1 A company can report strong results and still fall sharply if the guidance narrative fails to accelerate. The market has shifted from fundamentals-driven to guidance-event-driven, where each earnings release functions as a binary outcome with asymmetric downside.
The next phase of earnings season carries this risk forward. NVIDIA, AMD, and Marvell each face Q2 guidance events under the same conditions.1 Options markets are expected to price elevated implied volatility into these names ahead of each release. In-line guidance is now a downside scenario.
For AI infrastructure as a sector, the pattern raises a broader question about sustainability. Semiconductor stocks have served as proxies for AI spending confidence. When those proxies swing nearly 20% on guidance language, the market is signaling fragility in the consensus AI capex narrative.
The underlying spend is real. But the pace priced into valuations requires continuous upside surprises — a difficult standard to maintain across multiple guidance cycles, particularly as hyperscaler discussions grow more nuanced about deployment timelines and return on investment.
Q2 earnings season now tests whether the AI infrastructure trade can hold at current expectations, or whether the Broadcom move marks the first visible crack in a consensus that ran ahead of delivery.