Amazon CEO takes aim at Nvidia, Intel, Starlink and more in annual shareholder letter

Amazon CEO Andy Jassy’s annual shareholder letter reads something like a Kendrick Lamar score, if the rapper were an executive talking about the company and not a poetic Pulitzer Prize-winning musician.

Meaning you have to know the history to understand all the competitors Jassy is targeting, along with sweet personal stories about his unrealized dream of becoming a sportscaster and watching hockey games with his father.

Of course, Jassy doesn’t throw out the challenge directly. It takes a more nuanced approach. For example, in his challenge to Nvidia, he wrote: “We have a strong partnership with NVIDIA, and we will always have customers who choose to run NVIDIA” and we will always support these chips in their cloud.

But he also says: “Almost all AI to date has been implemented on NVIDIA chips, but a new shift has begun.” He says AWS customers “want better performance for the price,” which means Amazon’s locally manufactured Trainium AI chips.

Jassy says demand is so high for this chip that available capacity for the newest chip, the Trainium3, has almost run out. Remarkably, he says Trainium4 capacity has almost been sold out, which still has 18 months to go.

This means that Trainium has reached an annual revenue run rate of $20 billion. But if Amazon were a chip maker that sold its products to others, it would be worth $50 billion, he assumes.

Sure, Nvidia generated $215.9 billion in actual revenue last year. Nvidia may not be shaking in its boots just yet. However, Jassy presents Trainium as a tremendous up-and-comer.

Jassy did not spare Intel either. He points out that AWS’s Graviton CPU, which competes with Intel’s x86 architecture, is “now widely used by 98% of the top 1,000 EC2 customers,” aka some of the world’s largest companies. Two companies even asked to “buy out.” everyone “We can’t approve these requests due to the needs of other customers, but it gives you an idea of ​​demand,” he writes (emphasizing our Graviton instance capacity in 2026).

He promised that Amazon’s Starlink competitor, Amazon Leo, scheduled to launch in mid-2026, is already successful as well. It has won contracts from Delta Air Lines, AT&T, Vodafone, the Australian National Broadband Network, NASA, and others.

Interestingly, he also said that Amazon might look to sell robots one day. He wrote that it could turn all the data from the 1 million warehouse robots into “robotic solutions” for industrial and consumer uses. Is there a human Amazon in our future? We’ll see. He talked about Amazon’s other businesses as well, like same-day delivery, groceries, and drones.

But mostly, Jassy tried to validate the hundreds of billions of dollars in capital expenditures he had committed to. In February, he announced plans to spend $200 billion in 2026 on capital expenditures, mostly to build AWS data centers. That’s more than any of the other big tech companies, which also spend significant amounts on capital expenditures. Jassy’s offer to shareholders makes sense considering Amazon stock has fallen below $200 per share and has not recovered.

“We’re not investing nearly $200 billion in capex in 2026 based on a hunch,” he wrote, using as an example that his deal with OpenAI included the model maker pledging to spend $100 billion on AWS. Naturally, there are those who question OpenAI’s ability to deliver on all its spending promises.

Noting this, Jassy insists that other than OpenAI, “there are many other completed (and unannounced) customer agreements, or still in process,” that are lined up to purchase AWS capacity.

We’ll have to wait and see. Those who cause the bubble are never the ones who see (or acknowledge) its existence. “I have followed the public discussion about whether this technology is overrated, and whether we are in a bubble.” But he announced in this letter that this is not the case for Amazon, at least.