Amazon's Q2 Beats as AWS Growth Hits 37%, Its Fastest in 18 Quarters

Amazon's Q2 Beats as AWS Growth Hits 37%, Its Fastest in 18 Quarters

Amazon delivered the quarter Wall Street had been waiting more than four years for: its cloud engine roared back to life. Amazon Web Services grew 37% year over year in the second quarter of 2026 to $42.2 billion, its fastest expansion in 18 quarters and a decisive break from the deceleration that had shadowed the division since the last cloud boom. The result vaulted past analyst expectations of roughly $40.54 billion and blew through the 31% growth the Street had penciled in, sending a clear signal that the AI infrastructure cycle is now translating into hard cloud revenue.

The headline number sat inside a broadly strong report. Net sales rose 20% to $200.6 billion, up from $167.7 billion a year earlier and above consensus of about $196.47 billion, marking Amazon's first quarter above the $200 billion threshold for the April-to-June period. Operating income climbed to $27.5 billion from $19.2 billion in the second quarter of 2025. AWS alone contributed $16.6 billion of operating income, a margin of roughly 39.4% that underscored how profitable the cloud unit remains even as Amazon pours capital into new data centers and custom silicon.

"AWS is booming, growing 36.7% year-over-year in Q2 — our fastest growth in 18 quarters — and our AI and Chips businesses each eclipsed run rates of more than $25 billion," Chief Executive Andy Jassy said in a letter to shareholders. Both of those businesses, tied to Amazon's Bedrock AI services and its homegrown Trainium accelerators, are growing at triple-digit annual rates off that base. AWS as a whole is now running at a roughly $169 billion annualized revenue rate.

The Anthropic Wrinkle

The eye-catching profit line came with an asterisk. Amazon reported net income that included non-operating pre-tax income of $53.4 billion, driven primarily by a revaluation of its investment in AI lab Anthropic. That accounting gain — not cash earned from selling goods, ads or cloud capacity — helped push total pre-tax profit toward $62.6 billion. The markup reflects Anthropic's soaring valuation after the two companies struck a 10-year, roughly $100 billion arrangement in April to run Anthropic's large language models on AWS's Trainium chips. Investors will note the gain is a paper mark that can swing in either direction; the operating story underneath it is what matters for the durability of the beat.

That operating story is about backlog and capacity. AWS's contracted-but-not-yet-recognized backlog swelled to $496 billion during the quarter, a forward book of business that gives the division years of visibility. Jassy told analysts that leading AI labs Anthropic and OpenAI have each made multi-year, multi-gigawatt capacity commitments on Trainium, and that a widening roster of companies — from Uber and Pinterest to startups such as Poolside AI and Twelve Labs — are adopting the chips. He went further, framing AWS as a candidate to eventually become a $1 trillion annual revenue business.

To feed that demand, Amazon raised its full-year 2026 capital expenditure guidance to about $220 billion, up from roughly $200 billion previously. Management attributed a chunk of the increase to higher memory costs, a reminder that the AI buildout is colliding with tight supply chains for the high-bandwidth memory that trains and serves large models. For the current quarter, Amazon guided net sales of $197 billion to $202 billion.

Why It Matters

The reacceleration is the strongest evidence yet that AI spending has moved from experimental workloads to sustained, revenue-generating consumption at hyperscale. For much of 2024 and 2025, AWS grew more slowly than Microsoft Azure and Google Cloud, and skeptics argued Amazon had ceded the AI narrative to rivals with tighter model partnerships. A 37% print, paired with a nearly half-trillion-dollar backlog, reframes that debate. It suggests the constraint on cloud growth is no longer demand but capacity — the ability to stand up data centers, power and chips fast enough.

That is also the risk. A $220 billion capex year is a staggering bet that demand holds, and the memory-cost callout hints that input inflation could pressure margins even as revenue accelerates. Amazon's twin $25 billion run-rate businesses in AI services and custom chips validate its vertical-integration strategy: by designing Trainium in-house, it can offer AI compute at prices that pull labs and startups away from more expensive Nvidia-based capacity, protecting margins that a pure reseller could not. The Anthropic revaluation, meanwhile, illustrates how deeply the hyperscalers and model labs are now financially intertwined — a dynamic that boosts reported profits on the way up but ties Amazon's earnings volatility to the fortunes of a private AI company.

What to Watch

The key questions now are whether AWS can sustain mid-to-high-30s growth against tougher comparisons, and whether the record capex converts into revenue faster than it compresses free cash flow. Watch the trajectory of the $496 billion backlog for signs of how quickly contracted demand becomes recognized revenue, and watch memory and power costs for margin pressure. Investors should also scrutinize how much of future profit leans on Anthropic marks versus operating income — the cleaner the operating beat, the more durable the rerating. If Amazon can keep converting AI demand into cloud dollars at this pace, Jassy's trillion-dollar AWS may stop sounding like a stretch.