Amazon just pulled off a strange magic trick. Its cloud division, AWS, posted its best growth numbers in three years and added a massive amount of power capacity in a single quarter. The business is now running at a $142 billion annual pace. The reward for this performance was an immediate 10% drop in the company’s stock price.
Key Takeaways
- AWS recorded $35.6 billion in Q4 2025 revenue, a 24% year-on-year increase.
- AWS added over one gigawatt of power to its data center network in Q4.
- Amazon shares fell 10% in after-hours trading following the quarterly results.
The numbers are objectively huge. AWS brought in $35.6 billion in just the last three months of 2025. That is a 24% jump from the year before. For a business this large, growing that fast is rare. CEO Andy Jassy pointed out that growing 24% on a massive base is much harder than posting high percentages on a small one—a not-so-subtle jab at competitors like Microsoft and Google.
This growth was driven by new deals with heavy hitters like Salesforce, BlackRock, and the U.S. Air Force. It seems the rush to move data from on-premise basements to the cloud is speeding up again, fueled largely by the demand for AI.
The big deal
This report confirms that the AI boom is turning into actual infrastructure spending. Companies are not just talking about AI; they are paying Amazon to run it. AWS now makes up about 16% of Amazon’s total revenue, but it is the primary profit engine that keeps the rest of the company running smoothly.
The physical scale of this expansion is hard to grasp. In just three months, AWS added over one gigawatt of power capacity to its network. To put that in perspective, one gigawatt is roughly the output of a nuclear reactor. Amazon added that much capacity in a single quarter just to keep up with the demand for servers and chips.
How it works
Cloud computing allows companies to rent computing power instead of buying their own servers.
Think of it like a massive commercial kitchen. Instead of a restaurant building its own bakery, buying huge ovens, and paying for the electricity to run them, it simply rents space in Amazon’s facility. If the restaurant suddenly gets a huge dinner rush, they can instantly rent two more ovens for the night, then stop paying for them when the rush is over.
Amazon builds the data centers (the kitchen), pays for the electricity, and manages security. Companies upload their data and applications. Now, because AI models require massive computing power, those same companies are renting Amazon’s high-end chips to run their AI tools right where their data already sits.
The catch
Building these digital factories costs a fortune. The stock dropped 10% largely because investors looked at the bill. Amazon plans to increase its capital expenditures—spending cash on buildings and chips—which eats into short-term profits. The company missed Wall Street’s expectations for earnings per share, and the market punished them for it.
There is also a physical limit to this growth. Adding a gigawatt of power every quarter is an engineering nightmare. It puts immense strain on local power grids and requires finding energy sources that simply might not exist in certain regions. While the revenue is there, the cost to generate it is rising.
What now?
Amazon is betting that spending billions now to build AI infrastructure will pay off later, regardless of what the stock market thinks today. They are continuing to add capacity aggressively.
If you run a business on AWS, expect more AI tools to be integrated directly into your existing storage and database services. Amazon wants to make it impossible for you to leave by keeping your data and your AI models in the same place.
Watch the energy sector next. Amazon needs power, and they are adding it at a rate that will likely force new deals with utility companies or nuclear providers soon.
