You’ve probably noticed the absurdity of the modern tech gold rush. Everyone is scrambling to build the next great AI application, desperate to strike gold. But the real money has always been in the infrastructure.
So why is Amazon, the undisputed king of cloud compute, suddenly trying to pan for gold in the river with its new AI harness, KiroCrew?
The gold rush is never won by the guy selling the shovels—until he decides to drop them and start panning in the river.
Amazon Web Services is a behemoth because it provides the cheap, scalable compute that powers the entire internet. That is their moat. That is their most profitable, defensible position. Yet, instead of doubling down on being the absolute best pick-and-shovel provider for the AI revolution, they are trying to win in the application layer. It’s a strategic misstep that has smart observers pulling their hair out.
As one developer succinctly put it: “I’m not sure why Amazon is bothering here instead of just focusing on offering compute, which is where the money is.”
It’s a valid frustration. The application layer is a bloodbath. You have OpenAI, Anthropic, Google, and a thousand startups fighting for mindshare. Amazon’s defensible position isn’t building a slightly better chatbot or AI coding assistant; it’s making it dirt cheap for everyone else to build theirs.
You don’t win the cloud war by building a slightly better chatbot; you win it by making it dirt cheap for everyone else to build theirs.
When Amazon chases the application layer, they aren’t just wasting resources—they are leaving their flank exposed. AI-native companies are hungry, focused, and building infrastructure optimized entirely for AI workloads. If Amazon takes their eye off the compute layer, they might just find themselves outflanked by a leaner, meaner competitor who didn’t get distracted by shiny apps.
The paradox of Amazon’s AI strategy is that they already hold the winning hand, but they can’t resist folding to play a different game. KiroCrew might be a neat tool, but it’s a distraction from the goldmine they already own.
Amazon doesn’t need to be the brains of the AI revolution. It just needs to be the dirt the revolution is built on.
FAQ
Q: What exactly is KiroCrew?
A: It's Amazon's attempt to build an AI harness, essentially moving up the tech stack to compete in the application and tooling layer rather than just providing raw compute.
Q: Why is this a bad move for Amazon?
A: It distracts from AWS's core moat: cheap, scalable compute. The application layer is crowded and low-margin compared to infrastructure. Amazon should let others fight the app war while they sell the compute.
Q: What should Amazon be doing instead?
A: Doubling down on infrastructure. They need to be the ultimate pick-and-shovel provider, making it dirt cheap for AI-native companies to build their applications on AWS.