The editorial view: Amazon's cloud unit just posted its fastest quarterly growth in nearly five years, and the market's reaction is the real story. AWS revenue rose 39.3 percent to $46 billion in the third quarter, accelerating for a sixth consecutive quarter and reaching a $184 billion annualized run rate, while the company's backlog of signed commitments swelled to $578 billion and grew at a triple-digit rate year over year. Yet Amazon shares fell sharply after the earnings release on October 29, because investors were pricing something else: the cost of the AI buildout.
The numbers behind the acceleration
Amazon reported total revenue of $201.9 billion for the quarter, up 12 percent year over year, with operating income up about 61 percent to $28 billion and an operating margin of 13.9 percent, the highest on record. AWS operating income reached $17.8 billion on a 38.7 percent margin, up from 34.6 percent a year earlier. Management said the acceleration was driven by cloud and AI demand, and the company confirmed it signed major deals during the quarter with Meta and OpenAI, two of the largest potential AI workloads in the market.
The growth trajectory is unusually aggressive even by AWS standards. On the earnings call, chief executive Andy Jassy guided AWS to grow 40 percent in both 2027 and 2028, a two-year forecast that implies the cloud business could approach a trillion dollars in revenue within a decade. He told investors that there has never been a trillion-dollar revenue business, and that Amazon may have two of them in the next five to ten years.
Why the stock sold off anyway
The counterweight is capital expenditure. Amazon maintained its roughly $220 billion full-year capex outlook for 2026, aimed overwhelmingly at AI infrastructure, and free cash flow came in well under analyst expectations, with the company spending on data centers that are being built before the contracted workloads are fully powered on. Analysts noted that AWS is only about 23 percent of Amazon's total revenue, so a fast-growing but capital-hungry cloud unit is being weighed against a slower-growing retail base and a market that wants evidence of returns on AI spending, not just commitments.
Coverage of the results drew a sharp line between the operating story and the balance-sheet story: the backlog proves demand, but free cash flow is what pays for it. Until capacity bottlenecks ease and powered-on infrastructure starts converting backlog into margin, large infrastructure outlays are likely to keep weighing on cash flow and on the stock. That tension, not the growth rate itself, is what investors are trading.
The concern is industry-wide, not unique to Amazon. Microsoft and Alphabet both reported capital expenditure that fell short of estimates in their own quarterly updates, adding to broader worry about the cost of the AI buildout, while Microsoft moved to raise $25 billion in debt to fund its own AI expansion despite maintaining positive free cash flow. Analysts watching AWS infrastructure returns point out that management says payback periods on AI infrastructure are shortening, but also that AI margins remain below the mature core business; in other words, the economics improve with scale, but the scale itself is expensive to build first.
What this means for B2B buyers
For enterprises procuring cloud and AI services, the quarter contains several signals. First, the hyperscale market is consolidating around a handful of providers that can fund AI capacity at this scale, which narrows the list of vendors capable of hosting frontier workloads. Second, the Meta and OpenAI deals confirm that even the largest AI labs are pursuing multi-cloud strategies, renting capacity from multiple hyperscalers rather than locking themselves to one platform; buyers evaluating their own resilience should treat that pattern as the industry norm. Third, Amazon's plan to sell its Trainium AI chips to customers outside AWS, which surfaced in coverage of the results, points to a future where AI silicon procurement is decoupled from cloud procurement, giving infrastructure teams more negotiating room.
The strategic read is straightforward: cloud competition has shifted from price wars to an AI capacity race, and the winners will be the providers that can turn massive capital spending into operating leverage before the market's patience runs out. AWS is growing like a startup again, but it is funding that growth like a utility. Both statements are true, and both matter for anyone planning a long-term cloud commitment.
For enterprises, the practical checklist is about leverage rather than loyalty. Signed backlog in the hundreds of billions means committed pricing and capacity windows are being locked up years in advance, so buyers negotiating new cloud commitments should expect longer contracts and earlier capacity planning. The emergence of alternate AI silicon suppliers, from Trainium to competing accelerator roadmaps, gives procurement teams an opening to press for flexible pricing on compute rather than accepting pass-through costs. And the widening gap between the largest hyperscalers and everyone else means the decision of where to run AI workloads is now a strategic architecture choice, not a commodity purchase.
