Amazon's cloud division accelerated to its fastest growth rate in nearly five years during the third quarter, even as the company's expanding AI infrastructure program drove free cash flow deeply negative and sent its shares down more than 7% in after-hours trading at one point. AWS revenue rose 39.3% year over year to $45.984 billion, outpacing market expectations, while Amazon kept its 2026 cash capital-expenditure plan at roughly $220 billion.
Amazon reported total third-quarter revenue of $201.944 billion on Oct. 29, up 12% from a year earlier. Diluted earnings per share fell to $1.89 from $1.95, while both revenue and EPS came in below market expectations cited ahead of the release. The company forecast fourth-quarter revenue of $240 billion to $247 billion, placing the $243.5 billion midpoint modestly below the roughly $244.5 billion market estimate.
The results place Amazon among the large technology companies spending aggressively to secure data-center capacity, AI chips, and power infrastructure before demand for cloud computing fully materializes in reported revenue. AWS is already producing substantial profit, but its expansion now requires a level of upfront capital spending that is reshaping Amazon's cash-flow profile.
aws growth outpaces expectations
AWS operating income climbed about 55% from a year earlier to $17.773 billion, according to Amazon's earnings release. Its operating margin reached 38.7%, below 39.4% in the preceding quarter but sharply above 34.6% a year earlier.
The cloud unit's $45.984 billion in quarterly revenue implied an annualized run rate of about $184 billion. AWS grew 36.7% in the second quarter, making the third-quarter result a meaningful acceleration rather than a simple continuation of recent momentum.
Amazon also reported that AWS backlog — contracted cloud business that has not yet been recognized as revenue — rose to $578 billion at the end of September, from $496 billion at the end of the second quarter. The company cautioned that backlog does not translate directly into quarterly revenue because cloud-contract recognition depends on delivering computing resources and on customer usage.
AI services featured prominently in Amazon's explanation for the expanding demand. Spending by Fortune 500 companies using Amazon Bedrock rose nearly sevenfold year over year, while total token calls on the platform rose nearly 17-fold, Amazon said. Token calls measure requests processed by AI models and provide an indicator of usage, though Amazon did not disclose Bedrock revenue.
During the quarter, Bedrock added models from OpenAI, Anthropic, Moonshot AI, xAI, and Zhipu. Amazon said OpenAI models on the service received a built-in web-search feature. The company also introduced a public preview of Bedrock Managed Agents, designed to help developers run AI agents that can complete multi-step tasks.
Amazon said its Strands Harness tool can reduce task costs by 32% while maintaining comparable or better accuracy using the same frontier model. The company did not provide independent performance testing or revenue figures for the product.
data-center spending overwhelms cash flow gains
The cost of building capacity was visible in Amazon's cash-flow figures. Cash paid for property and equipment reached $77.259 billion in the third quarter, more than double the $35.095 billion recorded a year earlier. Net spending, after asset-sale proceeds and related incentives, was $75.878 billion.
Over the first nine months of 2026, Amazon paid $175.670 billion for property and equipment. Across the 12 months ending Sept. 30, operating cash flow rose 29% to $168.3 billion. Free cash flow moved to negative $42.357 billion, compared with positive free cash flow of $14.788 billion in the comparable prior-year period.
Amazon's own figures show the scale of the imbalance. Operating cash flow increased by $37.6 billion over the trailing 12 months, while net capital expenditures increased by $94.8 billion. The company is generating more cash from operations, but its infrastructure buildout is consuming it much faster.
Amazon Chief Executive Officer Andy Jassy said revenue commitments for Trainium and Graviton, Amazon's in-house chips, increased more than 20-fold year over year. The broader chip business posted triple-digit growth, he said. Trainium is designed for AI training and inference, while Graviton processors support a wider range of cloud computing workloads.
Amazon has previously said its internally developed chip business was operating at an annualized revenue run rate above $25 billion. The company plans to expand deployments of its next-generation chips and expects to broaden Trainium use outside AWS in the second half of 2027. Anthropic has used Trainium 2 chips for Project Rainier, a computing cluster Amazon previously said would contain more than 500,000 chips.
retail growth was affected by prime day timing
Amazon's retail businesses grew more slowly than AWS. North America revenue rose 6% to $112.852 billion, while international revenue gained 5% to $43.108 billion, or 8% excluding currency effects.
The company said the timing of Prime Day distorted the comparison. Most major markets held the event in the second quarter of 2026, while last year's full event occurred in the third quarter. Excluding the timing shift, Amazon said third-quarter revenue growth would have been about four percentage points higher.
Advertising revenue was not reported as a separate line item, though Amazon said advertising grew 12% excluding currency movements and forecast growth in the high teens for the fourth quarter. The company also plans to hire about 300,000 seasonal workers in the United States for the holiday period.
Operating profit for the group rose 61% to $28.076 billion and operating margin reached a record 13.9%. Net profit slipped 2% to $20.685 billion, largely because the previous year's quarter included $9.5 billion in pre-tax gains connected to Amazon's Anthropic investment.
cloud spending creates a more selective technology trade
Amazon's report offers a practical lens on the AI infrastructure cycle for cryptocurrency traders following technology equities and decentralized computing projects. Rising demand for computing capacity is measurable in AWS revenue, backlog, chip commitments, and capital spending. It does not automatically establish demand for token-based computing networks, whose prices and usage depend on their own economics, liquidity, token supply, and customer adoption.
The sharp after-hours stock move also reflected a familiar tension in AI-linked technology: traders rewarded AWS's growth acceleration, yet focused on the cost of sustaining it. Amazon's fourth-quarter operating-income range of $28.5 billion to $33.5 billion exceeded the $25 billion reported a year earlier, but the revenue outlook fell slightly short of expectations.
For digital-asset markets, broad claims that a decline in large technology stocks will necessarily trigger immediate crypto selling are not supported by Amazon's earnings alone. Bitcoin, Ethereum, and smaller tokens have each shown changing relationships with technology equities across different market periods. Amazon's results instead provide a clearer signal about the economics of AI infrastructure: cloud demand is expanding quickly, and the companies meeting it are committing unprecedented amounts of cash to data centers, servers, and proprietary chips.
Rising AI and cloud volatility? Learn how to navigate similar market swings with our guide on digital assets.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
