Amazon Has Become an E-commerce and Cloud Computing Behemoth. Now It's Set Its Sights on Another Billion-Dollar Market. And It Could Spell Trouble for Nvidia.

Amazon’s AWS AI chip business has reached a more than $20 billion annual revenue run rate, with Trainium2 nearly sold out and Trainium3 almost fully subscribed. The article argues this could pressure Nvidia to keep innovating, but it also says AWS is targeting lower-cost workloads rather than directly displacing Nvidia’s premium chips. Overall, the piece is constructive on Amazon’s AI opportunity and only a modest competitive caution for Nvidia.

Analysis

The market is still underestimating how AWS can use AI infrastructure to convert internal scale into an external pricing weapon. If Amazon monetizes Trainium at arm’s length, the real pressure on Nvidia is not immediate unit loss but a slower erosion of pricing power in the “good-enough” workload tier, where buyers care more about cluster economics and supply assurance than peak performance. That dynamic matters because it attacks the margin mix first: even modest share loss in non-premium deployments can force Nvidia to spend more on software, interconnect, and roadmap cadence just to defend its installed base.

The second-order winner is probably Amazon’s broader cloud flywheel, not the chip business itself. A credible lower-cost silicon stack makes AWS stickier for inference-heavy customers and AI startups that are cash constrained, which can pull more storage, networking, and managed services revenue through the platform. It also weakens the bargaining position of hyperscale peers and smaller neo-clouds that lack custom silicon economics, potentially widening AWS’s already strong cost gap over 12-24 months.

For Nvidia, the key risk is not a one-quarter headline but a multi-year compounding problem if Amazon proves it can ship racks externally with acceptable software support. The consensus is likely too complacent on timeline: these initiatives often start as internal cost saves and become full commercial products once utilization saturates, so the catalyst can arrive faster than the Street expects if Trainium2/3 supply stays tight. The main reversal would be a software moat shock in Nvidia’s favor—if CUDA and enterprise tooling keep expanding faster than any hardware price delta, Amazon’s chip push becomes a margin diluter for AWS rather than a true competitive threat.

Key Decisions for Investors

• Maintain a long AMZN / short NVDA pair for 6-12 months: Amazon captures the option value of externalizing AI silicon, while Nvidia faces gradual pricing pressure; risk is Nvidia software moat or a major new chip architecture reset.
• On NVDA, prefer downside hedges via 3-6 month put spreads rather than outright short: the stock can stay resilient on revenue momentum, but implied protection is attractive if AWS chip commercialization becomes a real narrative catalyst.
• Buy AMZN on pullbacks tied to broader tech weakness, targeting a 12-18 month horizon: the market is likely valuing AWS AI only as cloud expansion, not as a margin lever that can improve mix and customer retention.
• If Trainium externalization is confirmed, rotate a portion of NVDA exposure into AMZN call spreads: better asymmetric exposure to a re-rating of AWS as an AI infrastructure platform, with defined risk if adoption stalls.
• Avoid INTC as a sympathy trade: this is not a generic semiconductor share shift, but a platform-level procurement decision where software, distribution, and cloud control matter more than merchant silicon branding.