AWS will be a $1T business, Jassy says — but first, it will spend big

Amazon CEO Andy Jassy used the company's quarterly earnings call on Thursday to restate his ambition for AWS, the company's cloud division: he believes it can become the first business with $1 trillion in annual revenue. He framed the build as unprecedented and said the company is investing to stay the leader, with the goal of reaching that scale within the next five to 10 years.
Getting there requires continued heavy spending. Jassy said Amazon still expects cash capital expenditures of approximately $220 billion in 2026, the majority of it earmarked for AWS, and that the company expects to spend meaningfully more in 2027, while describing a clear line of sight to financial returns.
The current numbers give some support to the pitch. AWS revenue rose 39 percent year over year, which Jassy characterized as a trend of steady growth over a sixth consecutive quarter, and the business is running at a $184 billion annualized revenue run-rate. He forecast growth north of 40 percent in each of 2027 and 2028, citing continued innovation, new core and AI functionality, enterprises choosing AWS for both their core and AI workloads, and the volume of capacity the company adds each year. Jassy also said the enterprise is in the very early days of AI adoption and that the strategy is working so far.
Investor reaction was not enthusiastic. Wall Street analysts repeatedly pressed Jassy during the call to explain the scale of the spending, what signals suggest the growth will materialize, and what profit margins to expect. Shares were down about 6 percent in after-market trading on Thursday.
Jassy's answer leaned on timing: he called the AI boom a once-in-a-lifetime opportunity that justified a large commitment now, and said Amazon can fund the plan. He argued the investment would change the size and revenue of Amazon permanently.
For developers and product teams, cloud capacity is the practical variable underneath all of this. GPU instances and large-scale AI compute have been scarce and allocated on short notice across the major providers, and roughly $220 billion of planned 2026 capex, mostly for AWS, is the clearest signal yet of how much new data center capacity the company intends to bring online. Whether that capacity translates into easier access, more instance types or long-term reservation options is not something the call addressed, but the direction of the spending points that way; in practice, teams planning large training or inference workloads should still assume tight supply in the near term while buildouts catch up.
The broader pattern is not unique to Amazon. AI services sit on cloud infrastructure, so the same boom lifts demand for compute, storage and networking. Microsoft, one of AWS's closest competitors, recently reported another quarter of record revenue growth driven largely by its cloud business, and it too has spent heavily. This suggests the big cloud providers are competing on who can build capacity fastest rather than on price alone, a shift with real consequences for anyone budgeting multi-year infrastructure.
A likely trade-off is that aggressive capex depresses near-term margins and free cash flow even when revenue growth is strong, which is roughly the tension the analysts on the call were probing and which the after-market share decline reflects. Jassy's forecasts are also projections rather than commitments: growth north of 40 percent two and three years out depends on demand holding up, and a $1 trillion run-rate within five to 10 years implies sustained compounding from the current $184 billion annualized figure. Jassy's comments that the enterprise is in the early days of AI adoption is his characterization of where the market stands, and the demand it assumes has not yet shown up in the numbers.
Why it matters: For developers and product teams, AWS's spending plans hint at how much AI compute will be available and how quickly it arrives, and the pace of that buildout affects access to scarce GPU capacity and the pricing attached to it. Teams that depend on cloud AI infrastructure should read the capex guidance as a supply signal, and should treat Jassy's multi-year growth forecast as a projection, not a guarantee, given how quickly demand and pricing in this market can move. Inference: the near-term dynamic is likely to remain capacity-constrained even as billions of dollars of new infrastructure come online, because the same growth that justifies the spending also consumes it.
Based on reporting from the original publisher. Visit the source for full context and later updates.
Publisher excerpt
Amazon wants to be the king of AI cloud computing and become the first $1 trillion revenue business. But to do that it will have to continue on its capex spending spree, its CEO said Thursday.