NVDA remains the cleaner expression of the AI demand curve: this print reinforces pricing power and extends the window where earnings estimates keep moving up faster than the stock’s multiple can compress. The bigger second-order read-through is to the supply chain—if AWS is deepening its reliance on NVDA, then hyperscaler procurement is still more about securing scarce compute than optimizing unit economics, which is supportive for TSM and select networking/power names over the next 1-3 quarters.

For AMZN, the partnership is strategically positive but economically ambiguous near term. It strengthens AWS’s AI credibility, yet it also implies continuing capex intensity before revenue monetization catches up, so the market may reward the narrative while still discounting free-cash-flow pressure over the next 1-2 quarters. On the competitive side, this makes life harder for alternative accelerator efforts from AMD and custom silicon programs at other hyperscalers if the cloud buyer still prefers NVDA as the default.

The contrarian risk is that investors extrapolate a high-growth path too literally: long-dated revenue targets are less important than whether hyperscaler orders stay linear through the next two earnings cycles. If cloud spend pauses, lead times normalize, or export restrictions tighten, the multiple can de-rate even with strong reported revenue. The key falsifier is any sign that NVDA growth decelerates sharply below consensus over the next two quarters or that AWS capex guidance stops rising.