The market is treating this as an AI supply-chain stress test, not a clean earnings day. The key issue is whether inference demand remains centralized in merchant GPUs or starts migrating to custom silicon; if hyperscalers keep validating in-house chips, the first-order damage to NVDA is mix and pricing power, not total AI spend. That creates a cleaner relative beneficiary in AVGO, which monetizes customization, networking, and design-in content even if unit growth at NVDA stays strong.

For enterprise software, the setup is asymmetric. CRWD benefits if AI increases endpoint and identity complexity, because security budgets tend to rise faster than overall IT spend when attack surfaces expand; CRM has a harder proof point because it must show AI creates measurable monetization rather than just supporting retention. In the next 1-4 weeks, both names can be punished for simply meeting elevated expectations; over 6-18 months, CRWD should preserve a higher quality-growth multiple if it can keep converting deal flow.

META’s legal overhang looks more like a multiple-cleanup event than an earnings inflection. The contrarian risk is that investors are over-fixated on headline litigation relief while underweighting the real determinant for these large-cap duration names: whether elevated yields keep compressing long-duration software and internet multiples. Falsifier for the bearish AI-on-NVDA view is a print that shows stable gross margin, no meaningful custom-chip substitution, and unchanged hyperscaler demand trajectories; absent that, the stock can de-rate even on solid revenue.