AWS is printing money, but Amazon can’t stop spending it on data centers

Amazon reported its Q1 2026 earnings yesterday, and the headline is familiar: AWS is still the cash cow, but the company can’t seem to stop pouring money into infrastructure.

AWS revenue hit $28.8 billion for the quarter, up 22% year-over-year and beating analyst expectations by about a billion. That’s solid growth by any measure, especially when you consider the cloud market has been maturing for years now. But here’s the thing that caught my eye — Amazon’s capital expenditures hit $18.2 billion in Q1, and most of that went to AWS data centers and AI hardware.

CEO Andy Jassy was blunt on the earnings call: “We’re investing aggressively in AWS infrastructure, and we expect these investments to continue at elevated levels for the foreseeable future.” That’s corporate-speak for “we’re not done building yet.”

I get why. The AI boom is real, and it’s driving demand for compute power that didn’t exist two years ago. Every startup and enterprise is scrambling to train or deploy models, and they’re doing it on AWS, Azure, or GCP. Amazon is betting that if they build enough capacity now, they’ll capture that demand before competitors can scale up.

But $18.2 billion in a single quarter is a lot. That’s more than Amazon spent on all of its physical warehouses and logistics in any pre-pandemic year. The difference is that data centers depreciate slower than fulfillment centers, but they also lock you into a specific technology stack for years. If AI workloads shift to more efficient architectures or if demand softens, that’s a lot of sunk concrete and silicon.

The market didn’t seem bothered. Amazon’s stock was up about 3% in after-hours trading. Investors are buying the narrative that this spending is a moat, not a waste. And maybe they’re right — AWS still commands about 32% of the cloud market, and its operating margins, while compressed by the spending, are still healthy at around 28%.

What I don’t hear anyone talking about is what happens when the AI hype cycle inevitably cools. Cloud spending always follows a pattern: companies over-provision during boom times, then cut back hard when the economy tightens. AWS saw this in 2022-2023 when enterprises optimized their cloud bills. The same thing will happen again, probably within the next 18 months.

For now though, Amazon is riding the wave. Jassy mentioned that AWS signed several large multi-year deals during the quarter, including a $1.2 billion commitment from an unnamed financial services firm. That kind of locked-in revenue helps justify the spending, but it also means Amazon is betting that today’s AI workloads will still be relevant three years from now.

I’m not skeptical of AI’s long-term potential, but I’ve been around long enough to remember when “cloud” was the magic word that justified any expense. The companies that overspent then survived, but some of them had to write down assets. Amazon has the balance sheet to absorb mistakes, but that doesn’t mean they’re immune.

One more thing: Amazon’s overall revenue for Q1 was $158.4 billion, up 13% from last year. E-commerce is still growing, but the real story is AWS carrying the profit load. Without AWS, Amazon would be a low-margin retailer. With it, they’re printing money — and spending it just as fast.