Amazon's CEO Just Gave Nvidia Investors Great News

Amazon said its AI business is growing by triple-digit percentages, with chip revenue up 40% sequentially in Q1 and a deeper run-rate increase, while CEO Andy Jassy reaffirmed substantial purchases of Nvidia chips. The article argues this supports Nvidia demand even as Amazon expands its own Trainium, Graviton, and Nitro chips. For Nvidia, Wall Street is now looking for $79 billion in fiscal 2027 Q1 sales, implying 79% year-over-year growth, which raises expectations ahead of its May 20 results.

Analysis

The key takeaway is not that Amazon is “using less Nvidia,” but that its AI demand stack is expanding faster than its ability to self-provision compute. That creates a dual-benefit setup: Amazon monetizes AI as a platform layer while still acting as a large-scale buyer of premium accelerators, which should keep Nvidia’s utilization and pricing power supported longer than the market may expect. The second-order effect is that custom silicon is likely to suppress lower-end cloud economics before it meaningfully dents the high-performance tier where Nvidia still dominates.

For AMZN, the strategic win is margin mix, not just top-line growth. Every incremental workload migrated to Trainium/Graviton improves AWS unit economics, but the real upside comes from capturing budget-sensitive enterprise demand that might otherwise stay on the sidelines of AI adoption. That means AWS can widen the funnel while preserving the premium segment through Nvidia partnerships, a structure that should support share gains in the next 2-4 quarters.

For NVDA, the bull case is that concern around substitution is premature because the replacement cycle is asymmetric: custom chips can win on cost in constrained use cases, yet frontier model training and high-complexity inference still require the best silicon. The market is already pricing a strong growth path, so the risk is not demand collapse but expectation compression if capex rephasing or inventory digestion emerges over the next 1-2 quarters. META is a small indirect beneficiary as a large AI spender; LUV and NFLX are essentially noise here.

The contrarian read is that “custom chip vs Nvidia” is a false binary. The more AI workloads Amazon attracts, the larger the total compute pool becomes, and that expands Nvidia’s TAM rather than shrinking it in the near term. The trade implication is to own the platform winner and the picks-and-shovels winner simultaneously until evidence appears that custom ASIC adoption is displacing, rather than complementing, premium accelerator demand.

Key Decisions for Investors

• Long AMZN / long NVDA pair: hold for 3-6 months to capture expanding AI demand; thesis is that AWS self-supply boosts margins while Nvidia retains premium workload share. Risk: if AWS deployment shifts faster than expected to custom silicon, NVDA may underperform on sentiment even if fundamentals stay intact.
• Buy AMZN on pullbacks over the next 2-4 weeks, targeting a 10-15% upside over 6 months as AI mix improves and cloud demand broadens. Stop-loss below the recent post-earnings support zone if AWS commentary turns to capex discipline.
• Stay long NVDA into the next earnings window but hedge with short-dated calls or a collar into the print. Risk/reward is favorable for a beat-and-raise, but expectations are elevated, so the asymmetry is in protecting gains rather than adding aggressively pre-print.
• For more aggressive positioning, buy AMZN Jan-2026 calls financed by selling out-of-the-money NVDA calls. This expresses the view that Amazon’s operating leverage from AI monetization is underappreciated relative to Nvidia’s already-full multiple.
• Avoid adding to META, LUV, or NFLX on this headline alone; any benefit is second-order and likely already embedded in broader AI sentiment. Treat them as non-factors unless subsequent capex or partnership disclosures create a clearer earnings linkage.