Amazon CEO Andy Jassy’s annual letter to shareholders carries the subtle, pointed energy of a diss track—though one delivered in polished corporate language rather than poetic verse. To catch every competitive jab, you need the backstory, woven together with personal anecdotes about his childhood dream of sportscasting and watching hockey with his father.

Jassy never names rivals outright, opting for a more measured tone. Discussing Nvidia, he affirms, “We have a strong partnership with NVIDIA, will always have customers who choose to run NVIDIA” and will continue supporting those chips on AWS. Yet he quickly adds: “Virtually all AI thus far has been done on NVIDIA chips, but a new shift has started.” He notes that AWS clients are seeking “better price-performance,” pointing to Amazon’s in-house Trainium AI chips. Demand is so intense, Jassy reports, that capacity for the upcoming Trainium3 is nearly exhausted—and even Trainium4, still 18 months from launch, is almost fully reserved.

This surge has pushed Trainium to a $20 billion annual revenue run rate. Jassy speculates that if Amazon sold these chips externally, that figure could reach $50 billion. For context, Nvidia posted $215.9 billion in revenue last year, so it may not feel threatened just yet. Still, Jassy positions Trainium as a serious emerging contender.

EMBED_PLACEHOLDER_0

Intel also comes into focus. Jassy highlights that AWS’s custom Graviton CPU, an alternative to Intel’s x86 architecture, “is now used expansively by 98% of the top 1,000 EC2 customers”—many of the world’s largest corporations. Two firms even asked to “buy all of our Graviton instance capacity in 2026,” he writes, adding emphasis. “We can’t agree to these requests given other customers’ needs, but it gives you an idea of the demand.”

Looking ahead, Jassy shared updates on other ventures. Amazon’s satellite internet project, Project Kuiper, set to begin launches in mid-2026, has already secured contracts with Delta Airlines, AT&T, Vodafone, Australia’s National Broadband Network, and NASA. He also hinted at a future in robotics, suggesting data from Amazon’s one million warehouse robots could evolve into “robotics solutions” for industrial and consumer markets—leaving the door open to a potential Amazon humanoid robot down the line.

The letter touched on other areas like same-day delivery, groceries, and drones, but a central theme was justifying Amazon’s massive capital expenditures. In February, Jassy outlined plans to spend $200 billion in 2026, largely on expanding AWS data centers—a figure exceeding the capex of other major tech firms.

This emphasis makes sense given Amazon’s stock decline to under $200 per share, where it has remained. “We’re not investing approximately $200 billion in capex in 2026 on a hunch,” Jassy asserted, citing as an example that OpenAI committed to spending $100 billion on AWS. While some doubt OpenAI will fulfill all its spending pledges, Jassy countered that beyond OpenAI, “there are several other customer agreements completed (and unannounced), or deep in process,” ready to absorb AWS capacity.

Addressing broader industry skepticism, Jassy acknowledged the ongoing debate about AI hype and potential bubbles. “I’ve followed the public debate on whether this technology is over-hyped, whether we’re in ‘a bubble.’” In his view, at least for Amazon, that is not the case.