For years, Amazon basically told Wall Street to just sit tight. The spending was always massive, the timelines felt stretched thin, and the payoff was always promised for ‘some day.’ Trust the process, right? But Andy Jassy’s annual shareholder letter—dropped April 9—reads differently. He isn’t asking for patience anymore. He’s actually showing receipts.
It’s a shift. The market noticed, too, pushing the stock up as those numbers finally hit the screen. I remember checking the ticker while the smell of burnt coffee hung in our breakroom; the jump wasn’t just a blip, it felt like relief. Investors who were tired of vague, high-level AI talk finally got something concrete to sink their teeth into.
The big number? AWS’s AI revenue run rate is north of $15 billion as of Q1 2026. Misryoum noted this is the first time Amazon has actually pulled back the curtain on that figure. Considering the total AWS run rate sits at around $142 billion, AI is already doing the heavy lifting for about 10% of that business. It’s moving fast. Or maybe it’s just the scale that makes it look like it’s accelerating? Hard to say, but the demand is clearly eating up every bit of power capacity they can throw at it.
Then there’s the hardware. Jassy says the chip business—think Graviton, Trainium, and Nitro—is hauling in an annual run rate above $20 billion. That’s triple-digit growth. It doubled from the end of 2025. Honestly, the demand for Trainium chips is bordering on frantic; Trainium3 is basically sold out, and they’re already seeing reservations for the yet-to-be-released Trainium4. Jassy even dropped a hypothetical: if they sold these chips externally like Nvidia does, he claims the business would be running at $50 billion. That’s a bold claim, though I suppose it fits the mood.
There’s the $200 billion capital expenditure plan for 2026. A lot of money. Jassy insists it isn’t just a massive gamble on a hunch. He’s pointing to customer commitments—like that huge deal with OpenAI—to justify the spend. It sounds like he’s betting the farm on the idea that this is a once-in-a-lifetime window. If the future growth is as big as he says, well, then it’s not just a hunch, is it? Or maybe it is, just a very expensive one.
Analysts at Jefferies seem sold, keeping their $300 price target and calling these chips a ‘structural advantage.’ It’s interesting how sentiment shifts when the data actually arrives. Investors love clarity—or at least, they love proof that the money isn’t just vanishing into the ether. We’ll see how it plays out, but for now, the heavy spending cycle doesn’t look like it’s ending anytime soon.