Amazon Q1 2026 Earnings: AI, AWS, Guidance, And Monitoring Notes

Sources

Amazon Q1 2026 earnings release, released April 29, 2026.
Amazon Q1 2026 earnings call transcript, April 29, 2026.
Business Insider earnings recap, April 29, 2026.

Amazon reported a very strong Q1 2026, with revenue up 17%, operating income up 30%, and AWS growth accelerating to 28%, its fastest growth rate in 15 quarters. Management framed the quarter around three themes that matter for monitoring Amazon's AI execution: AI is accelerating AWS demand, Amazon's custom silicon business is becoming material, and the company is intentionally absorbing elevated CapEx and near-term free-cash-flow pressure to build long-lived AI infrastructure.

The most important AI/AWS signals were AWS reaching a $150 billion annualized revenue run rate, AWS AI services exceeding a $15 billion annualized revenue run rate, AWS backlog reaching $364 billion before including a new Anthropic deal worth over $100 billion, and Trainium revenue commitments exceeding $225 billion. Management also highlighted rapid adoption of Bedrock, OpenAI model availability in Bedrock, agentic AI products, Rufus, AI advertising tools, and internal AI productivity gains.

Amazon provided formal Q2 2026 guidance but did not provide updated full-year 2026 revenue or operating-income guidance. External reporting noted that Amazon had previously said in February that it planned roughly $200 billion of 2026 CapEx, much of it focused on AI infrastructure, and that Amazon did not update that figure on this call.

Key Financial Metrics

Consolidated Q1 2026 Results

Metric	Q1 2026	Q1 2025	Change
Net sales	$181.5B	$155.7B	+17%
Net sales excluding FX	Not disclosed as a dollar amount	Not disclosed	+15%
Favorable FX impact	$2.9B	N/A	N/A
Operating income	$23.9B	$18.4B	+30%
Operating margin	13.1%	11.8%	+130 bps
Net income	$30.3B	$17.1B	+77%
Diluted EPS	$2.78	$1.59	+75%
Q1 operating cash flow	$26.0B	$17.0B	+53%
Q1 purchases of property and equipment	$44.2B	$25.0B	+77%
Q1 cash CapEx cited by CFO	$43.2B	N/A	N/A

Important note: Q1 2026 net income included $16.8 billion of pre-tax gains in non-operating income from Amazon's Anthropic investment. That gain materially boosted GAAP net income and EPS.

Segment Results

Segment	Q1 2026 Revenue	Y/Y Growth	Q1 2026 Operating Income	Operating Margin
North America	$104.1B	+12%	$8.3B	7.9%
International	$39.8B	+19% reported, +11% ex-FX	$1.4B	3.6%
AWS	$37.6B	+28%	$14.2B	37.7%
Consolidated	$181.5B	+17%	$23.9B	13.1%

AWS was the standout segment. Revenue increased $2.0 billion sequentially from Q4 2025, which management said was the largest Q4-to-Q1 AWS revenue increase in AWS history.

Revenue Category Metrics

Category	Q1 2026 Revenue	Y/Y Growth
Online stores	$64.3B	+12% reported, +9% ex-FX
Physical stores	$5.8B	+5% reported, +4% ex-FX
Third-party seller services	$41.6B	+14% reported, +12% ex-FX
Advertising services	$17.2B	+24% reported, +22% ex-FX
Subscription services	$13.4B	+15% reported, +12% ex-FX
AWS	$37.6B	+28%
Other	$1.6B	+25%

Cash Flow And Capital Intensity

Metric	TTM Ended Q1 2026	TTM Ended Q1 2025	Change
Operating cash flow	$148.5B	$113.9B	+30%
Purchases of property and equipment, net of proceeds and incentives	$147.3B	$88.0B	+67%
Free cash flow	$1.2B	$25.9B	-95%

This is one of the most important monitoring areas. Amazon's reported earnings and AWS growth are strong, but free cash flow is being compressed by AI and AWS infrastructure investment. Management explicitly said that when AWS growth is very high, CapEx growth can outpace revenue growth and challenge early-year free cash flow until new capacity begins generating revenue.

Guidance

Formal Q2 2026 Guidance

Amazon guided for:

Net sales of $194.0 billion to $199.0 billion, representing 16% to 19% year-over-year growth.
Approximately 10 basis points of unfavorable FX impact.
Operating income of $20.0 billion to $24.0 billion, compared with $19.2 billion in Q2 2025.
Prime Day occurring in Q2 2026 in most major geographies, including the U.S.
Prime Day occurring in Q3 2026 in Australia, Brazil, India, and Japan.

Management called out several Q2 operating-income factors:

Seasonal step-up in stock-based compensation due to Amazon's annual compensation cycle.
Approximately $1.0 billion of year-over-year cost increase in North America related to Amazon LEO satellite manufacturing and launches.
Higher transportation costs from fuel inflation.
Partial offset from a recently implemented fuel and logistics FBA surcharge.

Full-Year 2026 Guidance

Amazon did not provide formal full-year 2026 revenue guidance or formal full-year 2026 operating-income guidance in the Q1 earnings release or call.

The company also did not provide fresh full-year CapEx guidance on the Q1 call. Business Insider reported that Amazon had previously said in February 2026 that it planned to spend about $200 billion on CapEx for the year, much of it focused on AI infrastructure, and that the company did not update that figure in the Q1 report or call.

Directional 2026 Commentary

Although Amazon did not provide full-year financial guidance, management did provide 2026 operating and investment milestones:

Cash CapEx in Q1 was $43.2 billion, primarily related to AWS and generative AI.
Amazon expects to continue making significant AI investments because management views AI as a massive opportunity for long-term revenue and free cash flow.
AWS CapEx intended for 2026 will partly be installed and monetized in future years.
AWS may spend cash 6 to 24 months before billing customers, depending on the component.
Data centers have 30-plus-year useful lives, while chips, servers, and networking gear have five- to six-year useful lives.
Trainium3 started shipping in 2026 and is nearly fully subscribed.
More than 1 million NVIDIA GPUs are planned to be deployed starting in 2026.
All new U.S. large-format fulfillment centers launching in 2026 will use Amazon's latest-generation robotics and automation technology.
Amazon LEO commercial service remains on track to launch in Q3 2026.

AWS Commentary

AWS reported $37.6 billion of Q1 revenue, up 28% year over year. Management said this was AWS's fastest growth rate in 15 quarters and that AWS is now a $150 billion annualized revenue run-rate business.

Management tied AWS growth to both core cloud services and AI services:

Customers are increasing cloud migrations and scaling AWS core services.
Customers pursuing AI are accelerating migration from on-premises infrastructure to the cloud.
Amazon sees a strong correlation between AI spend and core AWS growth.
As customers move more AI workloads into production, Amazon expects AI workloads to strengthen demand for core services.
AWS AI revenue is growing triple digits year over year.
AWS is bringing more capacity online to meet high demand while also improving efficiency across the installed base.

The key monitoring point is that Amazon is not presenting AI as isolated GPU rental demand. Management says AI demand also drives compute, storage, database, analytics, security, CPU, networking, post-training, reinforcement learning, agentic actions, tool usage, and other core AWS consumption.

AWS Backlog And Visibility

AWS backlog was $364 billion at the end of Q1.

Management said that figure does not include the recently announced Anthropic deal worth over $100 billion. Andy Jassy also said the backlog has reasonable breadth and is not just one or two large customers.

This is important because it gives Amazon unusually strong visibility into future AWS revenue demand, especially as large AI labs and enterprises commit to multi-year compute consumption.

AI Revenue And Products

AWS AI Revenue

Amazon said AWS AI services have exceeded a $15 billion annualized revenue run rate within the first three years of the AI wave. Jassy compared that with AWS's original launch trajectory, noting that three years after AWS launched it had a $58 million revenue run rate. He framed AI's ramp as nearly 260 times larger.

Bedrock

Amazon Bedrock was one of the strongest product signals in the call.

Key metrics and commentary:

Bedrock is used by more than 125,000 customers.
Nearly 80% of Fortune 100 companies use Bedrock.
Bedrock customer spend grew 170% quarter over quarter.
Bedrock processed more tokens in Q1 2026 than in all prior years combined.
A consistent majority of Bedrock workloads run on Trainium.
Amazon said 2025 Trainium2 throughput improvements increased token throughput by 4x, translating directly into more customer-serving capacity.

OpenAI Models In Bedrock

Management called the addition of OpenAI models to Bedrock a "big deal" for customers and Amazon's business.

Details provided:

OpenAI's GPT-5.4 model was added to Bedrock around the time of the call.
GPT-5.5 was expected to be added within the next couple of weeks.
Amazon started previewing Bedrock managed agents powered by OpenAI.
Jassy said customers have wanted to consume OpenAI models in Bedrock because customers want model choice.
Amazon expects stateful model APIs and managed agents to be important for future agentic applications.

Jassy emphasized that future agentic applications need to store state, identity, conversation history, and actions, and need to reach tools and compute resources. Amazon believes Bedrock managed agents address this requirement in a differentiated way.

Bedrock AgentCore

Amazon announced new capabilities for Amazon Bedrock AgentCore, AWS's infrastructure for building secure, scalable agents.

Reported capabilities:

AWS Agent Registry for discovering, sharing, and reusing AI agents, tools, and agent skills across an enterprise.
A managed agent harness that lets customers define the model, tools, and instructions while AgentCore stitches together the infrastructure.
General availability of Policy in AgentCore to control what agents can and cannot do.
AgentCore Evaluations for automated quality assessment.
AgentCore is being used to deploy an agent as frequently as every 10 seconds.

Strands, Quro, Transform, Connect, And Qwik

Amazon described several AI and agentic products:

Strands has been downloaded more than 25 million times and downloads increased 3x quarter over quarter.
Quro developer usage more than doubled quarter over quarter.
Quro enterprise customer usage increased nearly 10x.
Transform has saved customers more than 1.56 million hours of manual effort in workload migration and modernization.
Qwik new customers grew more than 4x quarter over quarter.
Amazon announced v1 of the Qwik desktop app, which can query email, calendar, Slack, local files, and other applications; flag communications; retrieve and summarize information; make recommendations; compose and send communications; and create agents.

Custom Silicon And AI Infrastructure

Amazon's chips business, including Graviton, Trainium, and Nitro, exceeded a $20 billion annual revenue run rate and is growing triple-digit percentages year over year. Jassy said the chips business grew nearly 40% quarter over quarter in Q1.

Jassy said that if Amazon's chips business were standalone and sold chips produced this year to AWS and third parties like other chip companies do, it would represent a $50 billion annual revenue run rate. He also said Amazon believes its custom silicon business is now one of the top three data-center chip businesses in the world.

Trainium

Trainium is central to Amazon's AI infrastructure strategy.

Key details:

Trainium revenue commitments exceed $225 billion.
Trainium2 has about 30% better price performance than comparable GPUs.
Trainium2 is largely sold out.
Trainium3 started shipping in 2026.
Trainium3 is 30% to 40% more price performant than Trainium2.
Trainium3 is nearly fully subscribed.
Much of Trainium4, which is still about 18 months from broad availability, has already been reserved.
Amazon expects Trainium at scale to save tens of billions of dollars of CapEx each year and provide several hundred basis points of operating-margin advantage versus relying on other chips for inference.

External Trainium Rack Sales

Jassy said Amazon may sell full racks of Trainium chips to external providers within the next couple of years.

The key constraint is allocation. Amazon must decide how much Trainium to allocate to:

Existing AWS demand.
Customers consuming Trainium through AWS cloud infrastructure.
Potential rack sales outside AWS.

This is a major future product and revenue-monitoring item because it could turn Amazon's custom silicon from an internal AWS advantage into a more direct merchant semiconductor business.

Graviton And CPU Demand From AI

Management emphasized that AI is not only a GPU or accelerator story.

Key details:

Meta committed to deploy tens of millions of AWS Graviton cores for CPU-intensive workloads behind agentic AI.
Agentic AI workloads include real-time reasoning, code generation, learning, and multi-step task orchestration.
Jassy said these workloads drive massive CPU demand.
Graviton delivers up to 40% better price performance than x86 processors.
Graviton is used by 98% of the top 1,000 EC2 customers.

This matters because Amazon may benefit from AI growth through both Trainium and Graviton, not just traditional GPU capacity.

NVIDIA Relationship

Amazon continues to buy and deploy NVIDIA GPUs.

The release said Amazon has landed more than 2.1 million AI chips over the past 12 months, more than half of which were Trainium, and announced more than 1 million NVIDIA GPUs to be deployed starting in 2026.

Jassy said Amazon will continue to partner with NVIDIA for as long as he can foresee because customers want choice and many customers want to run NVIDIA on AWS.

AI Datacenter Growth And CapEx Logic

Management's AI datacenter framework:

The faster AWS grows, the more near-term CapEx Amazon must spend.
AWS pays for land, power, buildings, chips, servers, and networking gear before the capacity is monetized.
The cash outlay usually occurs 6 to 24 months before customer billing begins, depending on the component.
Data centers can have 30-plus-year useful lives.
Chips, servers, and networking equipment can have five- to six-year useful lives.
Management expects compelling operating margins and ROIC once the assets are in service for a couple of years.
Near-term free cash flow can be pressured when CapEx growth exceeds revenue growth.

This is the central trade-off for investors monitoring Amazon's AI build-out: the company is creating large future capacity and backlog visibility, but free cash flow may remain depressed while infrastructure is being built ahead of monetization.

AI Partnerships And Commitments

Amazon highlighted several AI-related customer and partner relationships:

OpenAI committed to consume approximately 2 GW of Trainium capacity through AWS infrastructure for frontier models and advanced workloads, ramping in 2027.
Anthropic will secure up to 5 GW of current and future generations of Amazon Trainium chips to train and power advanced AI models.
Anthropic's new deal is worth over $100 billion and is not included in the $364 billion AWS backlog figure.
Meta agreed to deploy tens of millions of Graviton cores for agentic AI workloads.
Uber is using Graviton4 for ride and delivery matching and Trainium3 to train AI models.
Cerebras is collaborating with Amazon to deliver very fast AI inference speeds for large language models through Bedrock.
Other announced AWS agreements included NVIDIA, U.S. Bank, Fox, Southwest Airlines, U.S. Army, Bloomberg, AT&T, Nokia, Fundamental, National Geographic Society, PGA Tour, and others.

AI In Stores And Commerce

Rufus

Rufus is Amazon's agentic AI shopping assistant.

Key metrics and product details:

Monthly active users increased more than 115% year over year.
Engagement increased nearly 400% year over year.
Rufus can research products, track prices, and auto-buy products in Amazon's store when they reach a set price.
Jassy said Amazon is trying to make Rufus the best shopping assistant anywhere.

On third-party horizontal shopping agents, Jassy said the current experience is still small relative to search-engine referrals and is not yet great. He said third-party agents often struggle with pricing, product information, personalization data, and shopping history. Amazon is talking with these companies, but Jassy argued that a retailer's own agent can be more useful because it has better product information, personalization, account, and shipping context.

Agentic Commerce And Advertising

Jassy said he believes agentic commerce will be good for Amazon advertising.

His reasoning:

Agentic shopping experiences are multi-turn conversations rather than one-shot searches.
Multi-turn conversations create multiple opportunities to surface relevant products.
Some surfaced products will be organic and some will be sponsored.
Sponsored prompts inside Rufus are an example of this format.
Nearly 20% of shoppers who interact with a brand prompt in Rufus continue the conversation about that brand.

Amazon Ads generated $17.2 billion of Q1 revenue, up 22% year over year on an FX-neutral basis.

AI Advertising Tools

Amazon expanded CreativeAgent, an agentic tool that plans and executes the ad creative process, to Canada, France, Germany, India, Italy, Spain, and the U.K.

Jassy said AI tools reduce the time and cost required for small and medium-sized businesses to build creative and choose audiences. He expects AI to increase the number of advertisers.

Seller AI

Amazon introduced a new AI experience in Seller Central that dynamically generates personalized visualizations, insights, and scenarios tailored to seller goals. Management said the response is early but very strong.

Health AI

Amazon launched Health AI, a 24/7 AI-powered personal health agent backed by One Medical clinicians. It provides U.S. customers with clinical guidance and can take permitted actions such as booking appointments, managing prescriptions, and facilitating treatment with a One Medical provider.

Internal AI And Efficiency Commentary

Management said AI will have a "giant impact" across every Amazon business and every way Amazon works.

Jassy said Amazon is reassessing every customer experience as if it were being built from scratch with AI available. He expects many customer interfaces and interaction models to be reinvented over the next three to five years, possibly sooner.

Internal productivity examples:

Agentic coding is changing product development.
Jassy expects comparable changes in DevOps, customer service, research, analytics, and sales.
One internal service rebuild that would normally have taken 40 to 50 people about a year was completed by five AI-forward employees in 65 days using agentic coding tools.

This is important for monitoring Amazon's "AI efficiency" story: management is not just selling AI externally through AWS; it is actively applying AI to Amazon's own engineering, operations, and customer experiences.

Fulfillment, Robotics, And Operating Efficiency

Stores unit growth was 15% year over year, the highest since the tail end of COVID lockdowns. Fulfillment expense grew 9% year over year on an FX-neutral basis, while outbound shipping costs grew 12%, meaning unit growth outpaced key fulfillment cost growth.

Management highlighted:

Continued inventory-placement optimization.
Reduced distance traveled.
Fewer touches per package.
Better consolidation rates.
Robotics and automation as long-running productivity drivers.
Latest-generation robotics and automation being deployed in both new and existing facilities.
All U.S. large-format fulfillment center launches in 2026 will include the latest-generation automation technology.
Early positive results include improved site safety, higher productivity, and lower cost to serve.

This matters because Amazon's AI investment thesis is partly AWS revenue growth and partly internal cost-to-serve improvement across a huge fulfillment network.

Retail And Prime Metrics Relevant To AI Monitoring

Notable operating metrics:

Amazon delivered more than 1 billion items same-day or overnight so far in 2026.
More than 90,000 items are available for one-hour or three-hour delivery.
One-hour delivery is available in hundreds of cities and towns.
Three-hour delivery is available in more than 2,000 cities and towns.
Amazon Now offers 30-minute-or-less delivery on thousands of items.
Amazon Now began in India, where orders are increasing 25% month over month.
Prime members triple their shopping frequency once they start using Amazon Now.
Perishable grocery sales grew more than 40x year over year.
Perishables make up nine of the top 10 most ordered items for same-day delivery where available.
Same-day perishable customers add nearly 3x as many items to orders and spend over 80% more than customers who do not buy same-day perishables.
Amazon said it is now the second-largest grocer in the U.S., with more than $150 billion in gross grocery sales in 2025.

These metrics are relevant because AI shopping, fast delivery, grocery, and logistics efficiency can reinforce each other if Amazon successfully uses agents to drive conversion and fulfillment automation to lower cost.

Amazon LEO Relevance

Amazon LEO is not the main AI story, but it matters for CapEx, guidance, and long-term AWS integration.

Key details:

Commercial service is on track to launch in Q3 2026.
Amazon has more than 250 satellites in space.
Management said there are over 20 launches planned in 2026 and over 30 launches planned in 2027.
Jassy said LEO could become a "many-billion-dollar revenue business."
Management expects LEO to have AWS-like characteristics: capital intensive upfront, but with assets that can be leveraged over a long period.
Enterprises and governments want to move satellite data into the cloud, store it, analyze it, and run AI on it.
Amazon sees the combination of LEO plus AWS as compelling.
Q2 guidance includes roughly $1.0 billion of year-over-year North America cost increase related to LEO.
Amazon expects to begin capitalizing certain LEO costs in Q4.

LEO should be tracked because it adds near-term operating cost pressure, long-term infrastructure optionality, and potential AWS/AI data gravity.

Risks And Watch Items

CapEx And Free Cash Flow

The biggest investor concern is whether AI CapEx turns into high-return revenue and free cash flow. TTM free cash flow fell to $1.2 billion, down 95% year over year, because property and equipment investment increased sharply.

Monitor:

Quarterly cash CapEx.
Purchases of property and equipment.
TTM free cash flow.
AWS revenue growth relative to CapEx growth.
Timing between AI infrastructure spend and revenue conversion.
Whether Trainium produces the promised CapEx and margin advantages.

Memory And Supply Chain

Jassy said memory costs have "skyrocketed" and that there is not enough capacity for demand. Amazon began working with strategic suppliers in the middle to latter part of 2025 and believes it has done a good job avoiding capacity constraints, but it is watching the situation closely.

This can cut both ways:

Risk: memory and storage inflation can increase CapEx and pressure margins.
Opportunity: supply scarcity may push enterprises away from on-premises infrastructure and toward hyperscale cloud providers like AWS, which have better supplier access.

AI Demand Concentration

AWS backlog has breadth, but large AI labs are spending enormous amounts on compute. Monitor whether demand remains broad across enterprises, governments, and startups, or becomes too concentrated in a few frontier labs.

Agentic Commerce

Amazon believes Rufus and retailer-native agents can be strong, but third-party horizontal agents remain a strategic uncertainty. Monitor whether Amazon becomes the default fulfillment and product-data layer for external agents, or whether agents start routing commerce around Amazon.

LEO Cost Burden

LEO is a potentially large long-term business, but it adds near-term cost. Q2 includes roughly $1.0 billion of year-over-year cost increase related to LEO. Monitor launch cadence, service launch timing, customer commitments, and when costs shift toward capitalization.

Monitoring Checklist

AWS revenue growth rate and sequential growth.
AWS operating margin while AI capacity ramps.
AWS backlog, especially whether the Anthropic deal enters disclosed backlog.
AI services revenue run rate, currently over $15 billion.
Bedrock customer count, Fortune 100 penetration, spend growth, and token volume.
Trainium revenue commitments, currently over $225 billion.
Trainium2, Trainium3, and Trainium4 availability and subscription levels.
Evidence that Trainium lowers CapEx and improves AWS margin by several hundred basis points.
OpenAI and Anthropic capacity ramp timing, especially OpenAI's 2 GW commitment beginning in 2027 and Anthropic's up-to-5 GW commitment.
Whether Amazon starts selling Trainium racks externally within the next couple of years.
GPU deployment cadence, including more than 1 million NVIDIA GPUs starting in 2026.
Graviton adoption in AI-driven CPU workloads.
Rufus monthly active users, engagement, sponsored prompts, conversion, and auto-buy usage.
Advertising growth from agentic commerce and AI creative tools.
Internal AI productivity examples and whether they translate into lower operating cost.
Robotics deployment in new and existing fulfillment centers.
Unit growth versus fulfillment and outbound shipping cost growth.
TTM free cash flow versus CapEx.
Memory and storage supply cost trends.
Amazon LEO launch schedule, commercial launch in Q3 2026, and revenue commitments.

Bottom Line

Q1 2026 strengthened the case that Amazon is turning AI into a multi-layered business driver: AWS infrastructure demand, custom silicon, Bedrock and agentic AI services, AI commerce through Rufus, AI advertising tools, and internal productivity gains. The quarter also made clear that the opportunity is capital intensive. The central question to monitor is whether Amazon can convert today's AI infrastructure spending into high-margin AWS revenue, proprietary silicon advantages, stronger commerce conversion, and long-term free cash flow once new capacity is monetized.