Amazon’s 2025 Annual Report
Everything’s growing and Wall Street hates it
The disjunction between Wall Street reality and business reality was on display on the most recent Amazon quarterly earnings call, when Andy Jassy got up and said that everything is growing at a wild clip, and then Amazon’s stock promptly fell by 8%. The market reacted negatively at the announcement of up to $200 billion in capital expenditures, which is the first time in history that any company has announced that number (though Alphabet plans to spend up to $185 billion and Meta $135 billion—it’s all kind of insanity). Twice Jassy was asked if there were any capex guardrails, and his basic response was that this is a unique opportunity in a unique time, and they were going to uniquely spend a boatload of money.
Though the market reacted otherwise, all of this was said from a position of strength. Amazon announced revenue of a whopping $716.92 billion, which will probably be enough to finally dethrone Walmart as the company with the largest annual revenue. (We probably won’t have a Walmart 10-K until March.)
Over the last five years, they’ve had average annual revenue growth of 11.15%. If this continues, they will be the first ever company to make $1 trillion in a year by 2029. And if they achieve the 12.4% growth they managed last year moving forward, they’ll get there by 2028.
Operating income also rose from $68.59 billion in 2024 to $79.98 billion in 2025, after jumping up in 2024 from $36.9 billion in 2023. Again, these should be incredibly positive signs, but clearly we are in no ordinary time.
Revenue Segment Growth
I averaged out the growth rate of their key revenue segments, and the short of it is that everything is doing very well:
Bracketing the “Other” category (which includes “shipping services, healthcare services, and certain licensing and distribution of video content” - not sure why there would be such a marked jump between 2021 and 2022, as the One Medical acquisition did not close until 2023), one might be surprised to learn that the fastest growing segment of their business is not AWS but Ads. This is a concerning “Day Two” kind of trend in that it’s indication that Amazon is willing to degrade the user experience significantly in order to squeeze sponsored content in there, and one look at any Amazon search results page will tell you that they’ve lost the plot on that front.
It’s also something of a problem for their bet on their AI agent, Rufus. As with Google’s internal debate on whether or not to go all-in on Gemini or retain their typical search results page (with sponsored content), Amazon’s got to figure out how to monetize Rufus, as their in-house agent guides users away from all that sponsored content driving up ad revenue.
Physical Growth
Amazon announced that it now has 1.576 million employees, though this was before the recent round of predominantly corporate layoffs in January of 16,000 people. This would bring Amazon’s headcount down to 1.56 million employees, just above the 1.556 million they had in 2024. A few months back I tried to do a comprehensive assessment of Amazon’s automation game and concluded that no employment apocalypse was on the horizon at the company, even if displacement is happening all the time. I think this is probably what we’ll see from them going forward - incremental workforce increases that are not growing nearly as fast as revenue.
In terms of its physical footprint, Amazon continues to grow with 820 million square feet of space globally. Their Fulfillment, Data Center, and Other category is understandably the biggest one, and it’s notable that while that footprint is still expanding, it slightly slowed to 6.78% growth in 2025, after 9%+ expansion in 2022 and 2023.
Everyday Essentials and Grocery
One prominent and curious bragging point on the call concerned the growth in sales of “everyday essentials,” which supposedly now count for 1 in 3 Amazon purchases, as well as grocery. Jassy claimed that “we’ve become a go-to grocery destination for over 150 million Americans, mostly through online shopping and Whole Foods. With over $150 billion in gross sales, Amazon is clearly a large grocer at this point.” These are tremendously vague claims, and the fact that they don’t break out grocery anywhere on their 10-K is indication that this is more an indication of aims than an existing reality.
One might wonder why they even care about these low-margin categories when they’ve got this huge profit generator in AWS. I think the key reason is that these are both gateway categories: if someone’s buying everyday essentials and groceries from Amazon, chances are they’re buying a lot else on Amazon. And indeed, the company’s claimed that “customers who buy perishables through the offering shop twice as often as those who don’t.” As I will cover in a forthcoming column for Phenomenal World, Amazon’s Sub-Same Day centers are the key infrastructure for quick delivery of both everyday essentials and grocery, and so it’s no surprise that they are also among the fastest growing nodes in Amazon’s distribution network.
Vertical v. Horizontal Agents
Jason Goldberg latched on to a distinction Jassy made between “vertical” and “horizontal” AI agents as a key takeaway from the call. (Side note: Forbes’s ad use is truly inhuman. Reading that Goldberg article requires hunting for blocks of text around the images and video. Do better Forbes!)
Vertical agents are like Rufus: they’re retailer-deployed, they’ve got access to your full browsing and purchase history, and they’ve got pretty accurate product descriptions to work with. Horizontal agents are ChatGPT, Gemini, etc. - horizontal in the sense that their promise as commerce agents is working across retailers to find you the best deal. Jassy was quite critical of the latter on the call: “These horizontal agents don’t have any of your shopping history, they get a lot of the product details wrong. They get a lot of the pricing wrong.”
All of that’s true, but they’ll get better, and when they do, they’ll have an assortment rivaling Amazon’s and probably have a price advantage too. For this reason Goldberg thinks the horizontal agents are more of a threat than Jassy let on, and he takes Amazon’s block on 47 AI bots to be indication that they’re taking the threat seriously.
I don’t know what the agentic commerce future holds, but as long as Amazon has its distribution network, I’m not betting on any of the horizontal agents to pose much of a threat in the near future. Yes, horizontal agents can replicate the infinite catalog of Amazon, and yes, they can get a good price with cross-retailer comparison. But can they reliably get it to your doorstep same or next day? Goldberg recognizes that fulfillment is an advantage that vertical agents have, but I’m not sure it’s factored in to his analysis in the way it should be.
A Few Other Bits
Apparently Amazon has ten outbound regions now instead of eight?
The $200 billion in capex is mostly going to data centers, Project Leo, and Trainium chips, but there will still be substantial spending on the distribution network too.
There’s a $244 billion backlog for AWS services, i.e., “commitments in customer contracts for future services that we expect to fulfill but have not yet been recognized in our financial statements.”
If you’re wondering what the One Big Beautiful Bill did for Amazon, here it is: about $5 billion in federal tax savings.
