The cloud is growing faster than ever. And Amazon wants to know who will lead this revolution.
In the third quarter of 2026, AWS (Amazon Web Services) generated $45.98 billion in revenue — a 39% year-over-year increase. The figure beat Wall Street expectations, which had projected $45.29 billion. The division's operating profit reached $17.8 billion, up 55%, with a 38.7% margin. The growth rate is the fastest AWS has seen in nearly five years, surpassing competitors like Microsoft Azure and Google Cloud, which also accelerated but not at the same pace.
But the number that drew the most attention wasn't that one. During the investor call, CEO Andy Jassy made a statement that echoed across the entire market: AWS could become the first business in history to generate $1 trillion in annual revenue. This isn't distant speculation — it's a projection for the next 5 to 10 years.
"There has never been a business with a trillion dollars of annual revenue," Jassy said. "We may have two of them in the next 5 to 10 years."
He identified AWS and the retail segment (Stores) as those two potential businesses.
Jassy's confidence isn't unfounded. AWS has a contracted revenue backlog of $578 billion — more than double what it was a year ago. Major agreements with Meta, NVIDIA, OpenAI, Salesforce, Oracle, Qualcomm, Verizon, ADP, and Stripe are already signed. Tens of thousands of customers use AgentCore (AWS's AI agent platform), and they deployed 2x more agents in Q3 than in the previous three quarters combined.
Lambda, AWS's flagship serverless offering, just crossed $1 billion in annualized revenue. It's used by Samsung, Volkswagen, United Airlines, Mercedes-Benz, and Booking.com.
Now, there's a crucial detail that transforms this from "another good quarter" into a "potential inflection point": AWS capacity is essentially sold out through 2027.
"2027 is largely sold out," Jassy said on an a16z podcast. "A good chunk of 2028 is already booked." He added that Amazon "could invest a lot more and it wouldn't be a bad use of capital," considering 2026 capacity is already sold out.
This means something rare in technology: demand outstripping global infrastructure supply. When you're selling everything you can build, future growth is already locked in — at least until competitors can expand their own capacity.
The cost of this expansion is colossal. Amazon's capital expenditures (capex) in Q3 hit $77.3 billion — roughly $20 billion above expectations. For 2026, total capex will hover around $220 billion. And Jassy said explicitly that Amazon will spend "meaningfully more" in 2027.
Amazon's trailing-twelve-month free cash flow fell from an inflow of $14.8 billion in the same period of 2025 to an outflow of $42.4 billion in 2026. That's a swing of nearly $60 billion in a year. Most Wall Street analysts focused on that number, and Amazon shares fell more than 7% in after-hours trading following the results.
But Jassy has a simple argument: 85% of all global IT spending is still on-premises. He says we're in a 10-year cycle where that equation will flip to the cloud.
"Linchpin workloads, like mainframes and storage arrays, were never considered cloud-movable," Jassy said on the call. "But these workloads are moving. AWS has built cloud substitutes that make this possible."
The analogy is clear: it's the same reasoning that led companies to migrate from physical servers to traditional cloud in the 2010s, but now at exponentially greater scale, because the demand comes not just from conventional enterprises, but from AI labs training and running models with billions of parameters.
There's an interesting parallel with data center history. In 2008, when AWS launched EC2, many analysts doubted that third-party companies would trust critical infrastructure to the cloud. Over 15 years, cloud became the standard. Today, we're seeing the beginning of a similar movement, but toward an entirely new type of workload: inference and AI training at planetary scale.
AWS infrastructure also depends heavily on its own chips. Trainium, designed by Amazon, is used extensively for training AI models at lower costs than NVIDIA GPUs. Graviton, AWS's general-purpose chip, also contributes to margin. And the partnership with Anthropic — Amazon invested up to $25 billion in the company — is strategic: Anthropic will spend $100 billion on AWS in return, creating a trust loop between the two firms.
What differentiates AWS from the competition is vertical integration. While Microsoft relies heavily on existing Office 365 and Azure services to sell AI, and Google faces growing enterprise distrust after privacy scandals, Amazon has built an ecosystem that spans from custom chips to fulfillment, logistics, and now autonomous AI agents via AgentCore.
The risk, however, is real. The $220 billion in capex for 2026 is already the largest in corporate history. If AWS growth slows — from demand exhaustion, AI regulation, or any technological disruption — Amazon could have built too much capacity before revenue catches up.
But if Jassy is right about the 10-year cycle, then the current bet is only the beginning. The cloud computing sector is still in "uncharted territory," and whoever dominates AI infrastructure will likely define the next chapter of global computing the way AWS defined the last decade.
What will we say in 5 years, when AWS passes the $1 trillion mark? Will history repeat itself? Or will the next cycle have a different winner?
Sources: CNBC, TechCrunch, AI Weekly
✓ Independent sources cross-checked and verified before publishing
