AMZN | Weekly | Haru Insights

Amazon is committing to deploy 2 million Nvidia GPUs—Blackwell Ultra, Rubin, and Rubin Ultra—across its AWS infrastructure by 2028, signaling a firm strategic pivot toward full-stack AI dominance and cementing AWS as the backbone of next-generation AI workloads.
The upcoming finalization of the FAA’s BVLOS (Beyond Visual Line of Sight) rule is poised to unlock Amazon Prime Air drone delivery across seven new states, transforming the pilot program into a scalable, revenue-generating logistics network with major operational and competitive implications.

Amazon is constructing the world’s most advanced AI infrastructure at scale, leveraging Nvidia’s top-tier GPUs and custom Trainium chips—positioning AWS as the central nervous system of global AI and securing a first-mover advantage in enterprise AI deployment.
The company’s $220 billion 2026 capex plan, though massive, reflects a disciplined pivot toward long-term structural value creation in AI, energy (Project Prometheus), and autonomous mobility, extending Amazon’s dominance beyond retail and cloud into foundational tech ecosystems.
With Prime Air’s BVLOS rule nearing final approval, Amazon is on the verge of operationalizing a fully autonomous delivery network, turning a visionary initiative into a scalable, defensible logistics engine with potential for recurring revenue and competitive differentiation.
The integration of Nvidia’s Rubin and Rubin Ultra GPUs into AWS could give Amazon a performance edge over rivals, enabling faster model training and inference—key drivers in a market where speed and efficiency are paramount for AI leadership.
AWS’s vertical integration strategy, combining custom silicon, full-stack infrastructure, and exclusive partnerships, reflects a mature, defensible moat that could outlast short-term margins and entrench Amazon as the premier platform for AI innovation.

Amazon.com, Inc. (AMZN)[Tv]