re:Invent Keynote 2026: Analysis
Sticky Revenue, Unstoppable Growth, and the Behemoth That Is AWS
No one who attended the infamous re:Invent 2026 keynote would ever be able to forget the soul-searing horror they witnessed on the main stage that year.
AWS had just split off from the Amazon parent. As the CEO of the new standalone cloud computing company, Adam Selipsky faced high expectations for his first keynote address.
Some folks expected Selipsky to announce a radically new direction. Others expected a return to the Cambrian explosion of new service launches. The broader consensus was that AWS would continue with business as usual. Even Selipsky’s closest colleagues didn’t expect the 90-second performance that would come to be known as “The Keynote Incident.”
With thousands gathered live and even more viewing the keynote online, Selipsky walked onto the stage and calmly unveiled the Mona Lisa. He doused it in gasoline, set it on fire, turned, and left the stage without a word.
The world was stunned by the abhorrent destruction of Leonardo da Vinci’s masterpiece.
But nobody — absolutely NOBODY — expected that AWS would post its traditional 25% to 40% year-over-year growth rates the following quarter.
Don’t like AWS’s direction? Suck it up.
This is obviously a fanciful imagining of the future, and Selipsky doesn’t seem the type to set the Mona Lisa ablaze. But I believe it’s also an accurate depiction of what the business response would be to horrifying behavior from AWS.
If retail customers don’t like a company’s direction, they can cancel subscriptions or stop shopping at a store almost instantly. Advertisers can pause ad runs and cancel contracts remarkably quickly. But evacuating a cloud provider out of a distaste for its business practices, its leadership, or its CEO setting fire to a priceless cultural artifact live on stage is something that would take years.
Why you’re basically stuck with your cloud provider
American Founding Father and reputed insufferable jackass Ben Franklin once wrote that “three (moves) is as bad as a fire” for your household, and he wasn’t wrong.
There’s a reason that every significant cloud provider and also IBM effectively hurl money at customers to get them to attempt proofs of concept, migrate workloads, and pave the way toward cloud being as absolutely pain-free as possible: It’s incredibly difficult, time-consuming, and expensive to migrate workloads from one environment to another.
It’s why things like AWS’s Migration Acceleration Program shower customers with service credits and partner consulting hours to encourage migration. Once a workload lives on a given cloud provider, it’s extremely unlikely to be going anywhere anytime soon — and it sits there, generating revenue for the cloud provider every second that it’s running. There’s “sticky revenue,” and then there’s cloud revenue, which is something else entirely.
If you’re incensed by AWS’s ridiculous noncompete agreements or its [terrifyingly bad free tier](https://www.lastweekinaws.com/blog/an-aws-free-tier-bill-shock-your-next-steps/], you’re going to complain about it on Twitter or, if you’re me, on Last Week in AWS — but you’re not going to migrate your myriad complex workloads elsewhere without some seriously good incentives. You’ve got work to do, and “work” isn’t generally congruent with “endlessly shuffling your workload between providers.”
How growth endlessly bolsters AWS’s business
Most of the clients we work with at The Duckbill Group find that they’re spending more money on AWS a year after an engagement with us than they were when we started. It’s more efficient spend, and they can attribute it a lot better on a per-workload basis, but the raw dollars are higher. That represents growth. To be clear, it’s not a bad thing!
Holding an AWS bill at a relatively static level is hard. One of the few companies that has done this for more than a year is The Duckbill Group itself. That took some work to pull off, and we’re very good at it. For virtually every company that doesn’t have the capability and insight to drive work like that, the cloud bill is inherently an unbounded growth problem.
Those increasing cloud bills and workload growth means that if some regulatory or technical snafu were to suddenly decree that AWS could take on zero new customers for the next quarter or four, AWS’s revenue would continue to increase over that period.
Unlike the retail operation, the passions of the moment have remarkably little bearing on the cloud division’s business.
How the market ignores the AWS part of AMZN
I do not play the markets myself, but if I did, I would be absolutely gung-ho to invest in a pure cloud computing company. The near-guarantee of consistent growth and its lack of subjectivity to outside forces would make it a safe, profitable investment.
Unfortunately, we don’t have a pure cloud computing company that I could invest in among the big three. Instead, I would have to choose to invest in a bookstore, a software shop, or a surveillance company.
As it is, it’s frustrating to me to see Amazon’s stock price, earnings calls, and analyst forecasts basically ignore the incredible dynamics of the company’s cloud business in favor of talking endlessly about the vagaries of its retail business. The share price seems more driven by the commentary on Amazon’s retail results, not the cloud business.
But one of those businesses is incredibly variable and subject to the whims of customers, millions of disparate supply chain issues, and remarkably low margins.
The other is not.
As the industry matures, it’s hard not to see a dynamic unfolding where high-margin cloud divisions prop up low-margin ancillary businesses.
As an AWS customer, you’re basically stuck with the cloud provider you’ve already built your workloads on, no matter what Amazon might do or say.
As Amazon, you’ve nailed a business model that ensures consistent year-over-year growth from AWS, regardless of what you do or say.
As an investor, you’re looking for Amazon to split off the distracting retail business and let AWS thrive as its own company.
As an AWS employee, it doesn’t matter what you do: Your equity compensation structure is inextricably linked to how many underpants the retail store does or doesn’t sell.
As the Louvre, you’re keeping Selipsky far away from the Mona Lisa.