AWS by the Numbers: Q4 2025 and Full-Year Performance
Amazon.com (AMZN) reported strong results for its Amazon Web Services (AWS) segment in the fourth quarter of 2025, highlighting AWS’s importance to Amazon’s overall financial health. AWS generated revenue of $35.6 billion in Q4 2025, a 24% increase from $28.8 billion in Q4 2024 and a 7.8% sequential rise from Q3 2025. AWS accounted for roughly 17% of Amazon’s total Q4 2025 net sales of $213.4 billion but delivered approximately half of the company’s operating income, illustrating the segment’s profitability and strategic importance.
Operating income for AWS in Q4 2025 was $12.5 billion, up from $10.6 billion in Q4 2024 — a 17% year-over-year increase reflecting both revenue growth and efficiency improvements. For the full year 2025, AWS delivered approximately $128.7 billion in revenue, maintaining its status as the largest cloud services provider by revenue.
Metric
Q4 2025
Q4 2024
YoY Change
Full-Year 2025
AWS Revenue (GAAP)
$35.6 billion
$28.79 billion
+24%
$128.7 billion
AWS Operating Income (GAAP)
$12.5 billion
$10.6 billion
+17%
$45.6 billion
Amazon Total Net Sales (GAAP)
$213.4 billion
$187.8 billion
+14%
$716.9 billion
Amazon Net Income (GAAP)
$21.2 billion
$20.0 billion
+6%
$77.7 billion
AWS’s revenue growth outpaced Amazon’s overall sales expansion — a pattern that underscores the strategic importance of cloud and AI services within the company’s portfolio.
The AI Acceleration Inside AWS: Bedrock, Trainium, and Inferentia
AWS’s growth is closely tied to its investments in artificial intelligence infrastructure. Amazon Bedrock, launched for general availability in November 2023, is AWS’s managed service for building and scaling generative AI applications. Bedrock provides API-based access to a broad array of foundation models from providers including Anthropic, Meta, Cohere, Stability AI, and Amazon’s own Titan models — enabling enterprises to build AI applications without managing underlying infrastructure.
AWS’s AI infrastructure is further differentiated by its custom silicon. The Trainium3 chip is designed for efficient training of large AI models, offering customers improved price-performance for deep learning workloads. The Inferentia chip, optimized for inference, delivers low-latency, high-throughput model deployment at scale. Together, these chips underpin services like Bedrock and SageMaker, allowing customers to train and deploy large language models with improved energy efficiency compared to general-purpose GPUs.
By Q4 2025, Amazon’s custom chip business — inclusive of Graviton, Trainium, and Inferentia — had reached a greater than $10 billion annual revenue run rate, according to statements made on the Q4 2025 earnings call. Bedrock’s growing roster of enterprise customers spans industries from life sciences to financial services, reflecting broad adoption of generative AI workloads on AWS infrastructure.
How AWS Compares to Azure and Google Cloud
AWS remains the global leader in cloud infrastructure by revenue and operating income. In Q4 2025, AWS held approximately 28% global market share, ahead of Microsoft’s Intelligent Cloud segment (21%) and Google Cloud (14%), according to a report by CRN.
Provider
Q4 2025 Revenue
Market Share
YoY Growth
AWS
$35.6 billion
28%
+24%
Microsoft Azure*
Not disclosed
21%
29%
Google Cloud
$17.7 billion
14%
+48%
*Microsoft does not disclose Azure revenue separately; figures are estimates based on Microsoft’s Intelligent Cloud segment.
Google Cloud’s 48% growth rate outpaced AWS in Q4 2025, while Microsoft’s Intelligent Cloud grew at a comparable 29%. Despite faster competitor growth rates, AWS’s operating income of $12.5 billion in Q4 2025 exceeded both rivals — a function of its scale, pricing power, and proprietary infrastructure. In generative AI specifically, AWS’s multi-model strategy through Bedrock competes directly with Azure OpenAI Service and Google Vertex AI, with each platform emphasizing different strengths: Microsoft leverages OpenAI exclusivity and deep enterprise software integration, while Google Cloud emphasizes its Gemini models and data analytics heritage.
Capital Expenditure and the Infrastructure Bet Behind AWS AI
Amazon’s commitment to AI and cloud leadership is reflected in its capital expenditure plans. In its Q4 2025 earnings report, Amazon guided for approximately $200 billion in capital expenditures for 2026 — a significant increase from prior years — with the majority allocated to AWS data center expansion, networking infrastructure, and custom chip development.
This level of investment is designed to support the buildout of new AWS availability zones globally, expand capacity for Bedrock and AI services, and accelerate next-generation silicon development. By owning and operating its own chip design with Trainium2 and Inferentia, AWS aims to reduce reliance on third-party GPU vendors and optimize total cost of ownership for customers deploying generative AI at scale.
Amazon’s infrastructure strategy also includes geographic expansion for regulatory compliance: Bedrock is now available on AWS infrastructure in France and other European regions, addressing data residency requirements for enterprise customers subject to strict data governance rules. The magnitude of Amazon’s capital allocation to AWS signals management’s conviction that AI-driven cloud workloads represent a durable, multi-year growth opportunity — even as near-term free cash flow is affected by the investment pace.
Key Signals for Investors
AWS’s Q4 2025 operating income of $12.5 billion (+17% YoY) is expanding faster than revenue, suggesting that AI workloads are carrying higher margins and improving segment leverage as scale increases.
Amazon’s 2026 capex guidance of approximately $200 billion is among the largest infrastructure commitments in tech; investors should monitor whether revenue growth justifies the spend or whether return-on-capital timelines extend beyond current expectations.
The $10 billion-plus annualized run rate in custom chips (Trainium and Inferentia) indicates AWS is successfully monetizing proprietary silicon — a structural cost and performance advantage that Azure and Google Cloud do not replicate at equivalent scale.
Google Cloud’s 48% YoY growth rate in Q4 2025 outpaced AWS for the quarter; sustained outperformance from rivals would be a meaningful signal to watch in 2026 as enterprise AI budgets become more competitive.
Amazon’s full-year 2025 AWS revenue of $128.7 billion establishes a high base; maintaining double-digit growth rates from this scale will require continued enterprise AI adoption and successful expansion of Bedrock into new verticals and geographies.
