If I were to make a list of the least fashionable topics to talk about, it might start like this:

  1. Sewage systems
  2. EC2 instances

And those two things are more similar than you might think. Hear me out.

Sewage systems are critical; when a city’s sewage infrastructure stops working, buildings become uninhabitable, and civilization crumbles. Sewers are not exciting, nor are they something one discusses in polite company, but sewage disposal is the cornerstone of a functional society.

Cloud computing infrastructure is trending the way of the sewage system: It’s critical but blasé. It’s not exciting to talk about EC2 instances at great length in AWS keynotes anymore (with a notable exception for “Monday Night Live With Peter DeSantis,” which is actually “Surprise Computer Science Lecture With Professor DeSantis,” and it is absolutely glorious). Yet EC2 constitutes a majority of global cloud spend.

EC2’s fall into obscurity

Before AWS first launched, it was hard to get a server provisioned to build something. Virtual Private Server (VPS) offerings were as close as the industry really came, and the experience was lackluster at best. Suddenly, when AWS launched, you could provision the underlying resources to test an idea out in minutes rather than in months, and the investment to explore the concept was measured in pennies rather than tens of thousands of dollars.

Today, the world takes that for granted — and that’s OK. The hype cycle has significantly moved on, as has the state of the art. What gets the love and attention from the world is higher up the stack. It’s Machine Learning, not the GPUs that power it. It’s handling large-scale telemetry from IoT fleets, not hardening the computer on the manufacturing floor so dust doesn’t destroy it. It’s the latest serverless innovations, not the EC2 instances that are the lifeblood of most customers’ AWS estates.

This is a normal trend. A lot of Tier 1 backbone providers are absolutely critical to the functioning of the internet, but most internet users don’t realize that they exist in any meaningful sense. Like a municipal sewer system, the big pipes that move data around the world are critical infrastructure, but it’s not where the world’s attention is focused.

The future is higher up the stack

The start-up business of the future is almost certain to be radically different than it is now. Today, when someone graduates (or doesn’t graduate, as in my case) and has an idea to start a company that inherently requires software, they need to learn enough coding to build an MVP. That means they need to go to a boot camp, take a class, or struggle through with brute force and enthusiasm to articulate enough of a vision to raise funding to hire people who are good at computers.

It’s not always going to be that way.

In the future, those businesses aren’t going to buy a bunch of EC2 instances. Those entrepreneurs are going to turn to offerings that are higher up the stack — businesses that are AWS customers themselves but charge significantly more money than AWS, commensurate with the value they deliver.

For example, I pay $500 or so a month to AWS and receive about $500 of infrastructure services in return. I pay about $250 every month to Retool, incur probably 20¢ of infrastructure at absolute most, and receive enough business value in return that I’m absolutely thrilled to pony up. I have to imagine that AWS, as a vendor, would be absolutely thrilled to receive those kinds of margins on its services.

How we’ll pivot or become plumbers

It would be grand if AWS could create higher-up-the-stack products that deliver business value. The problem is that Amazon has spent nearly three decades being very, very bad at designing anything that even remotely looks like an appealing user interface. The front page of the retail web store isn’t a good site, it’s just that we as customers have grown used to navigating its non-ergonomic design. (Credit where it’s due: Amazon nailed the user interface for Kindle Oasis. Push the button, the page turns, and it gets out of your way.) Similarly, a building’s plumbing doesn’t have to be visually appealing or user-approachable, it just has to work.

But higher-order offerings that unlock greater business value — and their corresponding margins for vendors — absolutely need to get the customer experience correct. This is something that Microsoft understands in a way that AWS does not. Unless Amazon changes its trajectory, the future I see is that AWS goes the way of the Tier 1 backbone providers — effectively, a series of underlying infrastructure that powers the things that deliver customer value.

Folks who build on AWS now will either pivot to platforms that generate bigger business value or find themselves working in the province of plumbers