Amazon Q1 2026 Shows AWS Reaccelerating on AI Demand

Amazon’s first-quarter results show AWS growing at its fastest pace in 15 quarters, with AI infrastructure, custom silicon, and enterprise cloud driving a structural rerating of the business.

Key Takeaways

• AWS posted 28% revenue growth to $37.6 billion, its strongest quarterly rate in nearly four years, as enterprise AI adoption and custom silicon demand shifted from pipeline to revenue.
• Amazon’s custom chip business surpassed a $20 billion annualised run rate with triple-digit growth, while Bedrock token volumes in Q1 alone exceeded all prior periods combined, signalling AI monetisation at scale.
• With $200 billion in 2026 capital expenditure guidance and multi-gigawatt AI capacity commitments from Anthropic and OpenAI, Amazon is betting that infrastructure discipline today will yield dominant cloud economics tomorrow.

The Numbers That Matter

Amazon’s first-quarter results, released after market close on April 29, offered something increasingly rare in large-cap technology: genuine upside surprise at scale. Group net sales rose 17% year-over-year to $181.5 billion, group operating income expanded to $23.9 billion from $18.4 billion a year earlier, and net income reached $30.3 billion, or $2.78 per diluted share, against $1.59 in the comparable period. The headline numbers were strong. The details were stronger.

The net income figure carries an important footnote. A $16.8 billion pre-tax gain from Amazon’s investment in Anthropic inflated reported earnings materially; strip that out, and the operational picture is still compelling but less spectacular. What investors should focus on is the trajectory of operating income from the core businesses, and there the story is unambiguous. AWS, North America retail, and International each expanded margins in a quarter that saw the company maintain some of the most aggressive capital deployment in its history.

AWS: The Reacceleration That Investors Were Waiting For

The single most consequential data point in Amazon’s quarter was AWS revenue of $37.6 billion, up 28% year-over-year. That growth rate, building on 24% in Q4 2025, confirms that AWS has exited the post-pandemic digestion cycle and entered a new expansion phase driven almost entirely by AI-related workloads. Operating income for the segment reached $14.2 billion, making it by far the dominant profit engine of the group.

Context matters here. AWS is not a startup posting impressive growth from a modest base. It operates at a run rate exceeding $150 billion annually. Sustaining 28% growth at that scale requires either an extraordinary new demand vector or a fundamental shift in how enterprises are consuming cloud infrastructure. Both are present. AI has compressed the timeline for cloud migration decisions, and the complexity of training and deploying large models has concentrated spending on providers with the infrastructure depth to support them. AWS, with its combination of proprietary silicon, global data centre footprint, and managed service layers, is well positioned to capture that consolidation.

Custom Silicon and the Architecture Advantage

Perhaps the most strategically significant disclosure was the performance of Amazon’s custom silicon business. Graviton, Trainium, and Nitro collectively exceeded a $20 billion annualised revenue run rate in Q1, growing at triple-digit rates year-over-year. For a product line that barely registered in Amazon’s public reporting two years ago, this is a remarkable trajectory.

The commercial implications run deep. Custom silicon allows Amazon to offer performance-per-dollar economics that commodity GPU providers cannot easily match, and it creates a layer of architectural lock-in that strengthens long-term customer retention. The announced commitments from OpenAI for approximately two gigawatts of Trainium capacity, and from Anthropic for up to five gigawatts, are not merely revenue; they are multi-year structural endorsements of Amazon’s infrastructure thesis from the two AI companies that matter most to enterprise buyers.

Amazon Bedrock, the company’s managed model platform, processed more tokens in Q1 2026 than in all prior years combined. Customer spend on the platform rose 170% quarter-over-quarter. These are not vanity metrics. They reflect the transition from AI experimentation to production deployment among enterprise customers, a shift that is inherently stickier and more durable than the pilot phase that preceded it.

Retail: Quiet Efficiency, Real Progress

Against the noise of the AI narrative, Amazon’s retail business delivered results that deserve more credit than they typically receive. North America revenues grew 12% to $104.1 billion, with operating income rising to $8.3 billion from $5.8 billion a year earlier. International sales increased 19% to $39.8 billion, generating $1.4 billion in operating income. Unit growth reached 15%, the highest since the height of pandemic-era demand.

The margin improvement in retail is not accidental. It reflects years of investment in fulfilment automation, regionalisation of the logistics network, and the growing contribution of advertising, which now runs at more than $70 billion in trailing-twelve-month revenue and carries economics considerably more attractive than physical retail. Advertising’s continued high-teens-to-20% growth, powered by sponsored products, streaming inventory, and AI tools embedded in the shopping experience, gives Amazon a lever that most pure-play retailers cannot replicate.

Same-day and ultra-fast delivery expansion, along with Prime Day scheduled for the second quarter, maintain the flywheel dynamic that has long been Amazon’s competitive moat in commerce.

Capital Discipline at the Frontier

The one area where the market requires patience is capital expenditure. Trailing twelve-month capex rose sharply, and management has guided for approximately $200 billion in spending across 2026, the vast majority directed at AI data centre capacity. Free cash flow, as a result, contracted to $1.2 billion despite $148.5 billion in operating cash flow over the same trailing period, a compression that reflects the scale of infrastructure being deployed rather than any deterioration in underlying business quality.

The committed nature of much of this capacity is important context. When hyperscale customers sign multi-gigawatt agreements with defined utilisation schedules, the investment risk profile shifts from speculative to contracted. Amazon has been deliberate in communicating this, and the Anthropic and OpenAI commitments lend credibility to the utilisation outlook. The question of return timing is legitimate, but Amazon has navigated heavy investment cycles before and emerged with structurally improved unit economics on the other side.

Outlook and Positioning

Second-quarter guidance calls for net sales of $194.0 billion to $199.0 billion, representing 16% to 19% growth, and operating income of $20.0 billion to $24.0 billion. The range is wide enough to absorb currency volatility and Prime Day timing variability, but the midpoint implies continued momentum in both cloud and retail.

Amazon enters the remainder of 2026 occupying a position that few technology companies can claim: material exposure to each of the dominant secular themes in the current cycle, including generative AI infrastructure, enterprise cloud migration, programmatic advertising, and connected logistics. Its balance sheet is adequate to sustain the investment programme without compromising financial flexibility, and its competitive advantages in proprietary silicon, distribution density, and advertising data are not easily replicated.

The Q1 report is not the end of the story on AI monetisation, but it is meaningful evidence that the infrastructure being laid is beginning to generate the returns that justify the conviction behind it.