SEATTLE – Amazon Web Services posted its fastest revenue growth in roughly four years during the second quarter of 2026, driven by surging enterprise demand for AI infrastructure that pushed the cloud division past an annualized $100 billion revenue threshold and sent the parent company’s stock to its highest close of the year.
AWS grew 28% year-over-year to $29.3 billion in the quarter, its strongest expansion since the post-pandemic capex wave of early 2022. Chief Executive Andy Jassy said on an earnings call that AI-related workloads now represent an annualized run rate of $25 billion within AWS, a figure that has doubled in under twelve months as corporations accelerate their migration from traditional data centers to cloud-based AI infrastructure.
Amazon’s stock climbed more than 12% in after-hours trading after the results were published, recovering roughly two months of losses from a period when investors had grown skeptical that massive AI capital expenditure would translate into measurable revenue growth anytime soon.
The quarter resolves some of that doubt. AWS has seen its growth rate accelerate for three consecutive quarters, a trajectory that stands in contrast to the cooling pattern many analysts had projected following the spending frenzy of 2024. The unit generated $10.8 billion in operating income, its highest ever, and its operating margin expanded to 37%.
Jassy attributed the acceleration to three interlocking factors: a wave of enterprises moving production workloads to the cloud after years of pilot programs, a rush of AI-native startups choosing AWS as their primary infrastructure provider, and growing demand from government agencies for sovereign AI environments. The last category, he said, is growing faster than the others.
What Jassy declined to offer was a clear forecast for how long the acceleration can hold. Amazon’s capital expenditure for 2026 is now tracking above $100 billion for the full year, including aggressive investment in AI chips, custom silicon for both training and inference, and new data center capacity across three continents. The company lifted its guidance range for operating income in the third quarter while declining to break out AI-specific revenue beyond the run-rate figures it has disclosed in recent quarters.
That opacity has frustrated some analysts. Microsoft Azure grew 47% in the same quarter, with Azure AI services contributing an outsized share, and Reuters reported Google Cloud posted a 40% gain. Amazon’s 28% headline figure trails both, though AWS generates substantially more absolute revenue than either rival.
The earnings also underscored Amazon’s widening bet on custom silicon. The company’s Trainium 3 chip, designed for training large language models at scale, entered production during the quarter and is already in use by several large enterprises fine-tuning proprietary models on AWS. Amazon has invested $4 billion in Anthropic, whose Claude models are available through the Bedrock platform.
Jassy said demand for inference capacity, the compute required to run AI models against live user queries, was accelerating faster than training demand, which he described as a structural shift with multi-year tailwinds. The next inflection in enterprise AI adoption, he argued, will come not from experimentation but from production deployment, a phase in which companies integrate AI into core business systems rather than standalone applications.
Customers are still working through cost optimization as they scale AI workloads, Jassy acknowledged, but he framed that as a temporary phase. Enterprises that discipline their cloud spending tend to stay on the platform and ultimately expand usage as they identify higher-value applications.
Amazon’s overall revenue reached $187 billion in the quarter, up 11% from a year earlier, driven by advertising growth and a modest rebound in its third-party marketplace. Its advertising unit grew 19% to $17 billion. Meta’s second-quarter results showed a 22% advertising jump, putting pressure on Amazon to explain a deceleration in its ads growth, though currency headwinds account for part of the gap.
AWS remains the conglomerate’s financial engine. At a 37% operating margin, it generates more profit per dollar of revenue than any other Amazon division, and its trajectory through the second half of 2026 will largely determine whether the broader company can sustain the kind of shareholder returns that require multiple business units to perform simultaneously.
The risk lurking behind the headline numbers is that Amazon has committed to capital expenditure that rivals the annual infrastructure budgets of small nations. If enterprise AI adoption stalls or customers shift toward specialized models built atop rival silicon, AWS could find itself with expensive capacity and insufficient demand. That question is the part of the story this quarter cannot answer, and it is the reason investors will study the next two quarters with the same attention they gave the AI run rate this week.
