I'm LongbridgeAI. Here is the key information I extractedZhitong Finance APP learned that Amazon (AMZN.US) reported Q3 2026 revenue of $201.9 billion, a 12% year-over-year increase. Operating profit grew 61% to $28.1 billion. Chief Financial Officer Brian Olsavsky stated that the company's operating Profit Margin reached a record 13.9%. AWS revenue increased 39.3% year-over-year, with an annualized run rate of $184 billion and a backlog of $578 billion.
As AI and cloud demand drove accelerated performance growth, the company's cash capital expenditure in the third quarter reached $75.9 billion, with full-year 2026 expected to be approximately $220 billion. Expenditures are projected to rise significantly in 2027. During the earnings call, management stated that this infrastructure investment cycle aims to expand revenue, free cash flow, and return on invested capital in the long term, but noted that AI service margins are currently lower than the core cloud business and component prices remain a pressure point.
AWS Accelerates Growth Driven by AI and Core Cloud Demand
CEO Andy Jassy stated that performance was primarily driven by the continued acceleration of Amazon Web Services (AWS), growth in retail sales volume, and sustained investment in AI infrastructure. The company also forecasts Q4 net sales between $240 billion and $247 billion, with operating profit between $28.5 billion and $33.5 billion.
Specifically, AWS revenue totaled $46 billion this quarter, a 39.3% year-over-year increase and a $3.8 billion quarter-over-quarter increase. The cloud business's annualized revenue run rate reached $184 billion. AWS operating profit was $17.8 billion, with an operating margin of 38.7%.
Jassy noted that AWS growth has accelerated for six consecutive quarters, reaching its fastest pace in nearly five years. He mentioned strong demand for both traditional cloud computing and AI services, pointing out that AWS backlog reached $578 billion, representing triple-digit year-over-year growth.
"Almost as soon as we deploy compute capacity, customers snap it up," Jassy said, adding that Amazon deploys an AI accelerator approximately every 14 seconds.
The company stated that Amazon Bedrock—its managed service for accessing foundation models and building applications based on them—has become the fastest-growing billion-dollar service in AWS history. According to Jassy, spending by Fortune 500 companies on Bedrock increased nearly sevenfold year-over-year, and total token processing volume grew nearly 17-fold. Amazon recently added OpenAI models as well as hosted versions of Kimi, GLM, and SpaceX models to the platform.
Jassy also highlighted demand for the company's custom chips. He stated that revenue commitments for Graviton and Trainium grew more than 20 times year-over-year, and customers using Amazon's custom chips saved over $2 billion this quarter. Amazon plans to offer Trainium outside of AWS starting in late 2027.
Looking ahead, Jassy said Amazon expects AWS to grow more than 40% in both 2027 and 2028, believing the business could reach $1 trillion in annual revenue within five to ten years. He stated that this outlook reflects visibility into large-scale customer compute commitments, the ongoing migration of enterprises to the cloud, and the growth of AI inference in production use.
Capital Expenditure Rises with AI Infrastructure Construction
Amazon's total cash capital expenditure in the third quarter was $75.9 billion, primarily for AWS and AI infrastructure. The company continues to expect full-year 2026 cash capital expenditure to be approximately $220 billion, with the majority directed toward AWS. Jassy indicated that spending is expected to increase substantially in 2027.
Management views this investment cycle as a long-term opportunity to expand revenue, free cash flow, and return on invested capital. Jassy stated that the company is continuously shortening the time required for infrastructure investments to break even, though he noted differences in returns between core cloud and AI infrastructure.
Olsavsky said AWS profitability benefits from capacity optimization, fixed cost management, custom chips, and network investments. However, he pointed out that AI service margins are currently lower than AWS's core business, creating structural pressure on overall margins. Component prices also remain a headwind.
Jassy stated that Amazon owns or holds long-term leases for most of its data center capacity, combined with the use of custom chips, giving it a structural cost advantage compared to cloud providers more reliant on colocation or "neocloud" arrangements.
Retail Margins, Faster Delivery, and Grocery Expansion
Amazon's North America revenue grew 6% to $112.9 billion, while international revenue, excluding currency effects, grew 8% to $43.1 billion. Global paid units sold increased 12%.
Performance was impacted by the timing of Prime Day. This year, the promotional event took place in the second quarter in most major markets, whereas in 2025 it fell entirely in the third quarter. Olsavsky stated that excluding this timing shift, Q3 year-over-year revenue growth would have been approximately 400 basis points higher.
North America operating profit totaled $8.7 billion, with a margin of 7.7%; international operating profit was $1.6 billion, with a margin of 3.7%. Amazon stated that despite rising fuel and line-haul transportation costs, the company reduced overall fulfillment costs through optimized inventory placement, shorter shipping distances, and package consolidation.
Shipping costs increased 11% year-over-year, while unit volume grew 12%. Olsavsky stated that excluding fuel and line-haul inflation, shipping costs rose 8%, approximately 400 basis points lower than the unit growth rate.
Jassy said Amazon delivered over 1.5 billion items to Prime members via same-day or next-day delivery this quarter. The company's ultra-fast delivery service, Amazon Now, is now available in over 450 towns, with the number of items ordered through the service increasing tenfold year-over-year.
Grocery and daily essentials remain a focus. The growth rate for fresh grocery and daily essentials was approximately three times that of the rest of Amazon Stores, accounting for about half of this quarter's sales volume. Management stated that customers using same-day fresh grocery delivery purchase approximately seven times more daily essentials than those who do not.
Advertising, Pharmacy, and Q4 Outlook
Amazon Ads revenue in the third quarter was $19.6 billion, a 12% year-over-year increase. Jassy stated that growth was affected by the shift of Prime Day to the second quarter, the scheduling of NBA programming on Prime Video, and the comparison base from adjustments to search engine marketing projects in the same period last year. He said Amazon expects advertising revenue growth to rebound to the high teens in the fourth quarter.
Jassy stated that Amazon Pharmacy customers grew more than 60% year-over-year, with repeat customers increasing more than 50%. The company now offers three-hour prescription delivery in over 350 cities.
For the fourth quarter, Amazon expects exchange rates to reduce year-over-year net sales growth by approximately 60 basis points. Olsavsky also noted that fuel and line-haul transportation inflation in the fourth quarter is expected to be higher than in the third quarter.
Commercial services for Amazon Leo are still expected to begin initial operations before the end of the year. The company expects to begin capitalizing some Leo production and launch costs in the first quarter of 2027; therefore, most Leo-related costs are expected to continue being expensed in the fourth quarter.
Full Transcript Content
(I) Host (Operator)
Thank you for your patience. Hello everyone, and welcome to Amazon's Third Quarter 2026 Earnings Conference Call. (Operator instructions) This conference call is being recorded. For the opening remarks, please welcome Vice President of Investor Relations, Mr. Dave Fildes. Thank you, please go ahead.
(II) Dave Fildes – Vice President of Investor Relations Opening Remarks
Hello everyone, and welcome to our Third Quarter 2026 Earnings Conference Call. Joining us today to answer questions are our CEO Andy Jassy and CFO Brian Olsavsky. As you listen to today's call, we recommend having our press release handy, which contains this quarter's financial results, metrics, and commentary. Please note that unless otherwise stated, all comparisons in this call refer to the same period in 2025.
Our remarks and answers to questions represent management's views as of today, October 29, 2026, and include forward-looking statements. Actual results may differ materially. For more factors that may affect our financial performance, please refer to today's press release and our filings with the U.S. Securities and Exchange Commission (SEC), including our most recent Annual Report on Form 10-K and subsequent filings.
During this call, we may discuss certain non-GAAP financial measures. In our press release, the slides accompanying this webcast, and filings submitted to the SEC (all published on our Investor Relations website), you will find supplemental disclosures regarding these non-GAAP measures, including reconciliations of these measures to comparable GAAP measures. Our guidance incorporates order trends seen to date and assumptions we currently deem appropriate. Our performance is inherently unpredictable and may be materially affected by many factors, including fluctuations in foreign exchange rates and energy prices, changes in global economic and geopolitical conditions, tariffs and trade policies, resource and supply volatility (including memory chips), customer demand and spending (including the impact of recession fears), inflation, interest rates, regional labor market constraints, world events, growth rates in internet and online commerce, cloud services, and emerging technologies, as well as various factors detailed in our SEC filings.
Our guidance assumes, among other things, that we will not complete any additional business acquisitions, restructurings, or legal settlements. It is impossible to accurately predict demand for our goods and services, so our actual results may differ materially from guidance.
Now, I will turn it over to Andy.
(III) Andrew Jassy – President, CEO, and Director Keynote Address
Thank you, Dave. We reported revenue of $201.9 billion, a 12% year-over-year increase. Operating profit was $28.1 billion, a 61% year-over-year increase. The third quarter was another strong quarter for Amazon. On the AWS side, our plans are working. Despite adding significant capacity last year, AWS achieved 39.3% year-over-year growth, accelerating for the sixth consecutive quarter and marking the fastest growth rate in nearly five years. Backlog reached $578 billion, representing triple-digit year-over-year growth, and is highly diversified.
We install an AI accelerator approximately every 14 seconds, and customers consume them almost as soon as we install them. Our core compute and storage revenue posted the fastest growth rates in years. For a long time, our core business operating margins have been strong, and now AI business margins are improving rapidly, clearly ahead of where the core business was at a similar stage of development.
Despite rising costs for components like memory and significant depreciation from our capital expenditures, we continue to accelerate the payback on capital expenditures and are more confident than ever that our investments in AWS will generate revenue and free cash flow businesses much larger in scale than we could have imagined two or three years ago, with strong returns on invested capital.
We are forming a very attractive flywheel at AWS. Borrowing Jim Collins' definition of the flywheel from "Good to Great"—if you do A, you almost inevitably do B; if you do B, you almost inevitably do C, and so on.
At AWS, leading functionality, security, and operational performance inevitably drive customers to run applications and store data on AWS; this inevitably drives AWS AI demand (because customers want their production inference capabilities close to their other data and applications); this in turn drives core AWS demand (because AI needs to leverage core services for reinforcement learning, post-training, and tool calling); and this inevitably drives reinvestment in our broader AWS capabilities.
This is a different model from primarily hosting GPUs or short-term leasing data center capacity. It is a long-term platform strategy that broadly helps customers invent and evolve customer experiences, stabilize operations, and build reliable partnerships lasting years. At the same time, it delivers highly attractive long-term economics for us, regardless of whether supply is scarce.
There are many examples illustrating AWS progress; I will share just a few. On the core business side, it is important to remember that the vast majority (about 85%) of global IT spending is still on-premises, and we are in a 10-to-20-year cycle where this landscape will migrate to the cloud. Conventional wisdom held that critical workloads like mainframes or storage area networks would never migrate to the cloud. But these workloads are migrating because AWS has built unique cloud alternatives and AI-driven migration capabilities that make this transition easier. Enterprises gain lower capital expenditures, lower operating costs, faster innovation, higher developer productivity, and better data accessibility, making it easier to leverage AI.
Our core business margins are attractive because we have learned how to operate efficiently over the past 20 years. As the core business grows, we continue to reinvest in core AI capabilities. For example, as customers generate more data and seek value from analytics and AI, the complexity of moving data between storage and analytics engines is both burdensome and expensive. Through the acquisition of DuckLabs, DuckDB will become the universal SQL dialect for AWS storage and analytics, simplifying data processing with modern analytics engines—fast, easy to use, and ubiquitous. Customers need only write a SQL query once to execute it on any AWS analytics service.
In the analytics space, customers will for the first time achieve zero migration or zero portability costs between query engines. This is a big deal. You can also imagine this helping agents perform efficient queries.
On the AI front, we continue to expand the capabilities of our inference service, Amazon Bedrock, driving significant growth—Fortune 500 customer spending on Bedrock increased nearly 7-fold year-over-year, and overall Bedrock token volume grew nearly 17-fold year-over-year. Bedrock is the fastest-growing billion-dollar service in AWS history. We recently introduced OpenAI models to Bedrock, and usage is growing rapidly. We also just launched hosted versions of Kimi, GLM, and SpaceX models on Bedrock.
Today, you cannot talk about AI without talking about security. This is another reason customers continue to choose Bedrock. We designed Bedrock to ensure customer data remains secure and private. Inputs and outputs are never shared with model providers and are never used to train foundation models. When customers fine-tune foundation models, we create private copies exclusively for their use, with data fully encrypted both in transit and at rest.
As frontier models become increasingly powerful, they are discovering previously undetected software vulnerabilities at unprecedented speeds. We built AWS Continuum to discover, prioritize, validate, and fix code vulnerabilities at machine speed. Customers like Capital One, IBM, Infor, Luminate, and Robinhood are already adopting it. In the third quarter, we announced its integration into coding tools—Anthropic's Claude Code, OpenAI's Codex, and AWS's Kiro. Customer interest in this new security service exceeded anything we have seen before.
Agent deployment is proceeding at an accelerating pace, with many built on Bedrock AgentCore building blocks. Tens of thousands of customers currently use AgentCore, and the number of agents deployed in the third quarter was twice the sum of the previous three quarters. In addition to building their own dedicated agents, most companies also use out-of-the-box agent services, and we offer several beloved by customers. Kiro for coding, Amazon Connect for contact center agents, and Amazon Transform for migration. They are all iterating quickly and winning customer favor. So is Amazon Quick (our intelligent AI work companion), which is growing rapidly. Large enterprises like Mondelez, EchoStar, 3M, Southwest Airlines, and DXC choose Quick because it is built for work and makes no compromises on customer data.
Finally, our chip business continues to post triple-digit year-over-year growth, with revenue commitments for Graviton and Trainium each increasing more than 20 times year-over-year. AI continues to be a tailwind for Graviton, as post-training reinforcement learning and agent tool calling primarily run on CPUs (rather than AI accelerators), and Graviton is the most powerful CPU chip, offering 30% to 40% better price-performance than comparable products.
In the third quarter alone, customers using our custom chips saved over $2 billion. We also just open-sourced TorchNeuron, enabling developers to run their existing PyTorch code, which most are already using, on Trainium without rewriting any code. As we launch new generations of chips, demand for previous generations remains strong. After launching Graviton 5, backlog for previous-generation Graviton instances more than doubled quarter-over-quarter. Short-term capacity for previous-generation training instances is selling at a premium compared to the first half of the year.
Due to such strong training demand, customers continue to encourage us to offer this capability outside of AWS—we plan to achieve this by late 2027. Overall demand for AWS capacity remains exceptionally strong. Enterprises are still in the very early stages of scaling inference usage, yet inference already accounts for more than half of AWS AI revenue and is still growing. We also have substantial training revenue. We still expect 2026 cash capital expenditure to be approximately $220 billion, with the majority directed toward AWS, and anticipate investing even more in 2027. As I stated in the last conference call, we have a clear roadmap for substantial financial returns.
AWS is a business with an annualized revenue run rate of $184 billion, but the opportunity remains extraordinary. With our continuous pace of innovation, new core and AI capabilities, enterprises' desire to run their core and AI workloads on AWS, and the capacity we add annually, we believe AWS can achieve more than 40% growth each year in 2027 and 2028, becoming a $1 trillion revenue business within the next 5 to 10 years. We are in the very early stages of this inflection point, our strategy is working, and I expect it to continue working in the foreseeable future.
Regarding the Retail (Stores) business, I think some people underestimate the progress this business has made in recent years and its current scale. To my knowledge, no other retail business can grow at this speed while reaching Amazon's scale (hundreds of billions in revenue, with double-digit operating margins including advertising). Year-to-date, units sold increased nearly 15% year-over-year, built on the huge base of over 30 billion items sold last year. In absolute terms, we added over $90 billion in total sales in the past 12 months. Our growth rate is significantly faster than other retail competitors, adding roughly twice the total sales of our largest competitor in the US each quarter, and we continue to gain retail market share.
Over the past few years, our North America and International segments each expanded their operating margins by nearly 10 percentage points, allowing us to reinvest in better customer experiences. Remember, we are still in the early stages. The vast majority of retail (nearly 80%) still occurs in physical stores. Over time, this landscape will eventually shift to e-commerce. We are driving this shift by rapidly building an efficient, fast fulfillment network, delivering enjoyable customer experiences while generating substantial free cash flow and returns on invested capital.
From faster delivery to unit service costs, to groceries and pharmacy, to AI shopping, each of our priorities is a growth amplifier, and we are going all out. With our investments in same-day networks and ultra-fast services like Amazon Now and Amazon Prime Air, our delivery speeds continue to accelerate. We had the fastest third quarter in global history, delivering over 1.5 billion items to Prime customers via same-day or next-day delivery. We are encouraged by customer enthusiasm for the rapid global rollout of Amazon Now, which is now available in over 450 towns, with items ordered through Amazon Now increasing 10-fold year-over-year.
We are on track to lower unit service costs for the fourth consecutive year, allowing us to continue meeting or beating competitor prices and offering a wide variety of fresh food and daily essentials in a cost-efficient manner. Faster speeds are driving our grocery business, including fresh and daily essentials, with total sales growing to nearly $250 billion in the past 12 months. In the third quarter, fresh food and daily essentials grew at approximately three times the rate of other store businesses, accounting for about half of total units sold. Customers buying fresh food via same-day delivery purchase nearly 7 times more daily essential items than those who do not. Active grocery customers purchase non-grocery categories like home and apparel nearly 4 times more often than non-grocery customers, and spend more than twice as much in these categories.
Amazon Pharmacy is also growing rapidly. Customer count increased more than 60% year-over-year, with growth accelerating in each quarter of 2026, and repeat customers increasing more than 50% year-over-year. J.D. Power recently named Amazon Pharmacy the best mail-order pharmacy for the third consecutive year. We continue to improve, now offering 3-hour prescription delivery in over 350 cities.
In the past 12 months, over 400 million customers used our agent shopping assistant, Alexa for Shopping, with customer interactions increasing 570% year-over-year. US customers using Alexa for Shopping spend 50% more per order than those who do not. It researches products, compares options, reviews a full year of price history, provides personalized gift recommendations, and tracks deals and price drops for customers. Although we expect to collaborate with horizontal agents, we believe Alexa for Shopping will become the primary shopping agent customers use when shopping on Amazon.
Having a shopping assistant present while you browse, get inspired, and view product detail pages allows you to simply have it place orders based on past purchases—it already knows your history, shipping address, and payment information—offering a uniquely convenient experience for customers.
Turning to Amazon Advertising. We achieved revenue of $19.6 billion in the third quarter, a 12% year-over-year increase. Despite some timing anomalies last quarter, we are satisfied with the overall momentum of our advertising portfolio. I will mention three factors affecting growth. First, we moved Prime Day in most countries to the second quarter, shifting considerable sales and advertising revenue originally in Q3 2025 to Q2 2026—this was a tailwind for our Q2 2026 growth and a headwind for Q3 2026. For advertising, this had an 850 basis point sequential impact on Q3 year-over-year growth.
Second, Q2 2026 included the first season of NBA on Prime (which did not exist in Q2 last year), while Q3 did not include any NBA (NBA starts in Q4). Third, in Q3 2025, we adjusted our search engine marketing projects to drive more traffic to search pages on Amazon. This provided a better experience for customers, and we monetized this traffic at higher rates, boosting advertising revenue growth starting a year ago—we are now beginning to compare against that base. We expect Q4 year-over-year growth to return to the high teens.
A natural question is whether agent shopping from external horizontal agents is significantly impacting Amazon Advertising. The answer is: No. Very few customers are currently shopping heavily in this way because the experience is not yet good enough, nor is it more convenient than going directly to retailers to research or one-click ordering.
Our advertising team innovated fruitfully in Q3, with positive impacts expected in the coming months, led by Amazon Ads Agent—which integrates our complete advertising products under a single AI-first platform, supported by trillions of shopping, browsing, and streaming signals, providing advertisers with insights, answers, and the ability to optimize and act on their behalf at every step.
We also launched full-funnel campaigns in Amazon Ads Agent, integrating search ads, display ads, video, and streaming TV into a single campaign. Advertisers provide initial inputs (including budget and duration), and Amazon's AI handles planning, execution, and continuous optimization of the rest.
We continue to see momentum in several other areas. I will mention just a few. Prime Video has become a significant business, with a revenue run rate in the tens of billions and achieving profitability on its own. We have strong momentum with hit series like "Carrie," "The Summer I Turned Pretty," "The Boys," "Off-Campus," "Reacher," and "Fallout," blockbuster movies like "Project Hail Mary," and highly attractive live sports (including the best-starting Thursday Night Football of 2026, NBA on Prime, NASCAR, The Masters, and UEFA Champions League soccer).
We recently launched our most powerful device lineup to date, spanning 34 new products across Echo, Kindle, Alexa tablets, Fire TV, Ring, and Bee. It is still early, but we have started strongly and look forward to more customers getting their hands on these products during the holiday season. Finally, Amazon Leo remains on track to begin initial services this year.
We are entering the most exciting part of the year—with both re:Invent and the holiday season. I want to thank our teams for their hard work and innovation day after day, making customers' lives better and easier. The business we are building at Amazon is unprecedented. There has never been a business with $1 trillion in annual revenue. Amazon is on path to having two such businesses (AWS and Retail Stores). This is rare, and we are highly encouraged by what lies ahead.
I will now turn it over to Brian.
(IV) Brian Olsavsky – Senior Vice President and CFO Financial Review
Thank you, Andy. Let's start with overall revenue performance. Global revenue was $201.9 billion, a 12% year-over-year increase; excluding currency effects, it increased 13% year-over-year. Global operating profit was $28.1 billion, with an operating margin of 13.9%, marking our highest operating margin ever.
Let's look at North America and International segment performance. First, a reminder that Prime Day timing changed this year, with sales activities occurring in the second quarter in most of our largest countries, including the US. In 2025, Prime Day fell entirely in the third quarter. Excluding the impact of the Prime Day timing shift, Q3 2026 year-over-year growth would have been approximately 400 basis points higher. We saw strong customer engagement in Q3 and a continuation of H1 trends. North America segment Q3 revenue was $112.9 billion, a 6% year-over-year increase. International segment revenue was $43.1 billion, an 8% year-over-year increase excluding currency effects. Global paid units increased 12% year-over-year.
Prime remains a key pillar of our business. Hundreds of millions of Prime members are responding to our combination of vast selection, low prices, and record delivery speeds. Over the past year, Prime members more than doubled their monthly orders for fresh groceries with free same-day delivery.
Next, let's look at profitability. North America segment operating profit was $8.7 billion, with an operating margin of 7.7%. International segment operating profit was $1.6 billion, with an operating margin of 3.7%. Regarding the fulfillment network, the team managed rising transportation costs due to fuel inflation and higher line-haul rates. Despite these headwinds, we reduced overall unit service costs in Q3 through better inventory placement, shorter shipping distances, and package consolidation. Transportation costs increased 11% year-over-year, while units grew 12%. Excluding fuel and line-haul inflation, transportation costs grew 8%, approximately 400 basis points lower than the unit growth rate.
Robotics and AI are delivering measurable gains in productivity and employee safety, and we expect these gains to accumulate in the future. In 2026, we doubled our holdings of robotic arms like Cardinal and Sparrow. We are also developing robot foundation models to enable faster programming and teach robots increasingly complex physical tasks, aiming to scale the delivery of useful robots. Looking ahead, we see significant opportunities to improve productivity, lower unit service costs, and deliver products to customers faster across our global fulfillment network. Our operating margins will fluctuate quarter by quarter, but we focus on continuous long-term improvement in unit service costs.
Next, let's look at the AWS segment. Revenue was $46 billion, a 39.3% year-over-year increase, driven by both core services and AI services. We added $3.8 billion in revenue quarter-over-quarter. AWS currently has an annualized revenue run rate of $184 billion. Customers continue to migrate to the cloud and expand their use of AWS core services. Moreover, more customers seeking the full benefits of AI are accelerating their migration to the cloud.
We see a strong correlation between AI spending and our core service revenue growth. As customers invest in AI, we see corresponding increases in core consumption, and we expect this relationship to strengthen as more AI workloads move to production and drive new demand for compute, storage, and networking. AWS operating profit was $17.8 billion, with an operating margin of 38.7%, reflecting strong growth and our focus on efficiency improvements. Our investments in custom chips and custom networking, along with software and process improvements, are helping us optimize capacity and build more cost-effective infrastructure. As we always say, we expect AWS operating margins to fluctuate, partly depending on our investment levels at any given time. Our AI products are gaining customer traction, with AI revenue growing triple-digits year-over-year. Bedrock has been a significant growth driver.
Now, let's look at our cash capital expenditure, which was $75.9 billion in Q3. This is primarily related to AWS and AI, as we are investing to support strong customer demand. We continue to make large-scale investments, especially in AI, because we believe this is a huge opportunity likely to deliver long-term revenue, free cash flow, and returns on invested capital.
Finally, I will conclude with Q4 financial guidance. Q4 net sales are expected to be between $240 billion and $247 billion. Based on current exchange rates, we estimate the year-over-year impact of currency changes to be approximately a 60-basis-point headwind for this quarter. Q4 operating profit is expected to be between $28.5 billion and $33.5 billion. I will mention one factor affecting Q4 guidance.
Commercial services for Amazon Leo are scheduled to begin initial operations this year, and we expect to begin capitalizing some costs, including production and launch costs, in the first quarter of next year. This is later than my previous forecasts in prior earnings calls. Therefore, we will continue to expense most Leo-related costs in Q4. We are ready to serve customers this holiday season and support the millions of customers building on AWS.
I want to thank Amazon teams for their hard work and customer-centric focus. Our commitment to improving customer experience remains the only reliable path to creating long-term shareholder value.
We will now move to the Q&A session.
Q&A Session
(I) Q1 – Bank of America Securities Analyst Justin Post
Question: Andy, great to see cloud business continuing to accelerate, excellent. But sequential growth has slowed, I want to understand if there are factors on the customer side or token costs? And how do you view capacity additions in Q4? Also Brian, what is the impact of fuel costs on Q4 guidance?
Andrew Jassy Response:
Okay, I will answer the growth question first. We are very satisfied with this growth—a 39.3% growth rate for a business with an annualized revenue run rate of $184 billion is quite rare. This is our fastest growth in nearly five years, and the sixth consecutive quarter of acceleration. We are also adding significant capacity. I think if you look at the factors driving growth currently, it is a combination of several aspects. First, in AI, we have extraordinary full-stack capabilities—whether you are building your own models with SageMaker AI (the easiest way to build proprietary models) or want to leverage leading frontier models with high security and highly competitive price-performance, you can use Bedrock.
Whether you are building your own agents (we provide capabilities like Strands and AgentCore to build and run securely and scalably) or using out-of-the-box agents (the choice for most companies), just look at Kiro (coding), Connect (contact center), Transform (migration), and Quick (rapidly growing AI work companion), people are adopting these very quickly.
We add significant capacity every quarter. The amount of capacity added each quarter is not the same. But if you look at the capacity we have added this year, and the plans for the next ~8 quarters (we typically have visibility for 8 to 10 quarters), we have substantial capacity coming online. On top of that, add the impact of chips. We are in a very favorable position in the AI wave because the reality is that more and more customers—every customer using AI at scale, especially enterprises now starting to scale inference in production—are extremely eager to save costs on price-performance and training, and Trainium's significant price-performance advantage over other solutions is highly attractive.
As more people use AI in very broad ways, this drives demand for a large amount of core capabilities, such as reinforcement learning, post-training, and tool calling, and we have the leading CPU chip Graviton, which offers about 30% to 40% better price-performance than other x86 providers. These factors are truly driving very significant growth. The capacity we add each quarter will vary, but at this scale, predicting more than 40% year-over-year growth in 2027 and 2028 is significant, and we believe this can become a $1 trillion revenue business in 5 to 10 years, an extraordinary, rare opportunity that we are very excited to seize.
Brian Olsavsky Addition:
Justin, regarding fuel and line-haul inflation, I don't have exact numbers, but Q4 will be higher than Q3. Remember we launched a fuel surcharge for FBA sellers at the beginning of the year, recovering part (but not all) of the fuel costs we saw for these shipments and remaining partners in retail inbound and outbound freight. So it is included, slightly higher than Q3.
(II) Q2 – Morgan Stanley Analyst Brian Nowak
Question: I have two questions. Andy, your enthusiasm for AWS came through on the call. You rarely make statements like "$1 trillion revenue in 5 to 10 years," which is beyond the usual scope. Could you elaborate on how you reach that $1 trillion? Is it top-down or bottom-up? What gives you confidence in achieving such revenue growth and reaching this number in the next 5 to 10 years?
Second question, regarding robotics, looking at 2027, what are the key milestones you are focusing on for robotics in the retail business?
Andrew Jassy Response:
Regarding AWS, Brian, just look at the growth rate—look at the run rate growth rate, and then look at our forecasts for 2027 and 2028. As I have said many times, regardless of how much capacity we provide, the bulk capacity customers acquire (hundreds of megawatts, sometimes up to 1 gigawatt) is reserved in advance, because you cannot let 1 gigawatt of capacity sit idle. Therefore, we have very clear visibility into this bulk capacity demand for 2027, 2028, and part of 2029.
We will always reserve a significant portion of capacity for on-demand use on EC2 or Bedrock, and work closely with customers to ensure they have sufficient capacity available in AWS. But we have very clear visibility into what AWS will look like in the coming years, and when we calculate revenue against the existing run rate, such growth rates will quickly push the numbers up.
On top of that, I do not think much of the usage we see today will decay quickly. Enterprise deployment of inference in production is still in the very early stages, really early. Partly because enterprises need enough talent who know how to operate AI, perform post-training, and fine-tune models for suitable applications; partly because of inference costs. I think if we want enterprises to use inference as we believe they will, we must continue to strive to lower inference costs. This is a major mission for AWS, and one of the important reasons we build our own custom chips and Trainium—we firmly believe it is extremely important to allow customers to use inference more economically. Inference costs have dropped significantly in the past two years, and I think they will drop significantly in the future.
Then think about what you are seeing. Take security, for example. Just this alone, models are now so advanced that they discover vulnerabilities humans have never noticed. If you can use these models, you can use them to discover vulnerabilities and ultimately help patch and deploy them. Some people believe that a considerable portion of the total tokens consumed in the next year or two may ultimately be used for security, a relatively new use case,叠加 on top of the coding growth we are seeing and various AI companions (in consumer and workspace spaces, like Quick). These are just the first few large use cases.
Add to this the enterprise trends you will see—I believe almost all modern enterprise production applications will be rewritten to have agent capabilities, because if you want to move quickly, or want to guard against certain security risks in the right way, you must deploy this way, the AI wave is just too large. So, you cannot predict precisely in 5 to 10 years, especially with such growth rates, but for us, imagining this could become a business exceeding $1 trillion is not an exaggeration.
Regarding robotics, what I want to say is that what is happening in the robotics field currently is quite astonishing—overall, and especially at Amazon. We now have over 1 million robots in our fulfillment network. They have had a very significant impact on our cost structure, productivity, and safety—look at our fulfillment network safety data, which is at historic lows. We have made huge progress in the past four or five years, and robotics is an important part of that.
But I want to say that whether regarding what we have built and deployed, or what we are developing, we are still in the early stages. I think you will see that our fulfillment centers will always have a large number of employees. But I think much of the more repetitive work currently done by humans will be done by robots. We are investing in robots of various different forms. We also have opportunities to help complete the "last 100 yards" delivery to customer stores. We have many results coming soon.
For us, the metric for robotics is: does it bring equal or better productivity, safety, cost structure, and delivery times? When the economic combination is favorable, we will scale deployment in the fulfillment network. If you look at the new designs coming to our fulfillment network—including the expansion of same-day facilities—they are heavily invested in robotics. So we think robotics has already had a huge impact, and this is just the tip of the iceberg compared to the next five years.
(III) Q3 – Evercore ISI Analyst Mark Mahaney
Question: Two things, first thank you very much for the AWS growth outlook. Could you also provide an outlook on margins? I know margins fluctuate quarter by quarter, but over the next two years, do you think the current margin structure is relatively stable? Will we maintain AWS operating margins in the 30s or 40s?
Also regarding fresh grocery checkout, I am super excited to see this opportunity landing in my community and other areas. How extensive is your fresh grocery checkout coverage? In other words, how many Prime households do you think you cover? For customers who have used it, how much do you think this increases spending per Prime household? Is it a 20%, 50%, or 100% increase? Given that the fresh food category is so large?
Brian Olsavsky Response:
Mark, I will start with the AWS operating margin outlook. We do not provide guidance on AWS margins. But I can give you my perspective. This quarter's operating margin of 38.7% looks very strong. I want to add that the profitability you have seen recently is not a random result, but the product of disciplined efficiency improvements, capacity optimization, and strict control of fixed costs. As Andy said, over time, more custom chips are entering our cost structure. However, there are other factors at play. In fact, the extraordinary growth of AI puts pressure on overall business margins because margins there are lower.
We are satisfied with the progress in both core and AI margins. As Andy said earlier, they are ahead of where the core business was at a similar stage, but still lower. So there is this mix issue. There are also headwinds from component pricing we have discussed. So it is unpredictable, but there will always be multiple factors intertwining, and we work very hard on the cost and efficiency sides, while handling margins for new products—as they launch, release, and reach scale.
Andrew Jassy Addition:
Regarding fresh grocery checkout, Mark, I am as excited as you are. As you know, we have been doing large-scale grocery business for some time. It mainly focuses on middle-aisle items in a grocery store—daily essentials, canned foods, medicines, beauty products. We have the leading organic grocery store Whole Foods Market, which is also growing faster than other parts of the grocery business, both in revenue and profit, and the new format performs very well in urban environments for daily shopping, which we plan to roll out in the coming years.
We have always known that if we want to serve customers in the way we think they want, and we want, we must offer mass-market products, and this all starts with fresh food. If you look at total household spending, when we compare the total amount consumers and customers could spend with us versus what they actually spend with us, the biggest gap is groceries. And a lot of shopping, a lot of groceries, starts with fresh food. So, we are indeed very excited because we think we have found the right direction with delivering fresh food from same-day facilities.
Our daily essentials growth rate is already about 3 times that of other parts of the store, currently accounting for about half of our total units. We found that customers buying fresh food via same-day delivery purchase about 7 times more daily essential items than those who do not. So, if you are trying to narrow the gap between total consumer spending and the goods we can provide, fresh food has a huge impact on this gap.
We also see that our grocery customers purchase non-grocery categories like home and apparel about 4 times more often, and spend about 2 times more in these categories, compared to customers who do not buy groceries from us. These are huge differences, and we believe the progress we are making in fresh and grocery will ultimately make us more relevant to customers in more shopping scenarios.
(IV) Q4 – Goldman Sachs Analyst Eric Sheridan
Question: Two questions about the AWS market opportunity. First regarding Bedrock, what are your key insights into the Bedrock market opportunity? How will it continue to expand in the coming years? Second, regarding the decision to sell Trainium externally, how do you view the balance between the market opportunity for external sales and the compute needed for this opportunity, and its balance with internal use cases and other AWS opportunities?
Andrew Jassy Response:
Regarding Bedrock, you may have heard us say this many times in the past two years. But 20-plus years of experience providing technology to customers tells me that no single tool rules the world. People always think the early leader will be the only thing used, but that has never been the case. It was true for databases, and it was true for analytics. In the early days of AI, everyone said the only framework was TensorFlow, but now PyTorch is much larger.
It is the same for models. In the next 2 to 3 years, I think there will be 6 to 8 models that are comparable in level, or at least sufficient to cover more than 95% of workloads. Some are better at inference, some at coding, some at vision, some have lower latency, some are lower cost. Most customers will use multiple models, and many will use multiple models in the same generative AI application.
Therefore, what customers really want—something we have repeatedly verified in conversations with customers—is a service with a wide selection of these frontier models, good price-performance, and strong security and access controls. They want this service to decide what combination of models to use, and to switch back and forth as models surpass each other and application constraints change. This is exactly what we are seeing with Bedrock. People are very excited about this.
We have both closed-source models like Anthropic and OpenAI, and open-source models like Kimi and GLM. I think the Western world is very eager for more Western open-source models, and I believe this will happen over time. But I think the scale of the Bedrock opportunity is huge; it could become a service as large as EC2, and EC2 is now a very large service within AWS. So I think it is very large.
Regarding the balance of chips—how much to sell externally versus how much to use internally in the cloud, this is a good question and something we are continuously evaluating. I want to say that there are actually only two companies that have both a significant cloud business and chips sufficient to sell externally. So there are not many customers who really need to face this kind of problem at such scale. Currently, everyone is using the same foundry to produce their advanced chips. This foundry can only provide so many chips per year, and everyone gets a quota. You must decide how much of your quota to sell externally and how much to use for your own cloud.
There is a trade-off here. If you sell externally, you can realize more revenue in the short term, because you can recognize revenue as soon as the chips leave the warehouse. Conversely, if you decide to use them within the cloud in the form of EC2 instances or Bedrock, you actually drive more absolute revenue and free cash flow from these chips, because you can use them continuously for 5 to 6 years, generating revenue over a longer period. So, you really must think carefully about what to do. To date, I would say we have taken a different approach to chip usage than the other provider facing the same balance issue.
As I mentioned earlier, we believe selling chips externally will be a very substantial business, and we will launch it in late 2027. We may rethink this balance every year. But both will be very important. Our ability to build high price-performance, scalable chips will be a very significant strategic advantage at scale—both in our own cost structure and as an independent business selling chips externally—selling chips externally is indeed a business quite distinct from the cloud infrastructure business.
(V) Q5 – JPMorgan Analyst Douglas Anmuth
Question: When you consider AWS revenue growth exceeding 40% in the next two years, and the multi-year path thereafter, can you discuss the capital expenditure required to drive this revenue growth in the coming years and beyond? Also, Andy, do you have updated thoughts on the payback periods for servers, networking, and data centers?
Andrew Jassy Response:
Okay, I will answer in reverse order. Regarding infrastructure break-even, as I mentioned in the opening remarks, we are continuously moving it to the left—making it shorter. The core business and AI are slightly different, and financing methods also differ depending on which part you are talking about, but we are continuously making progress. As we have always been good at in AWS history, we constantly look for ways to extend useful life and accelerate break-even, and we are making good progress in this area.
Regarding the capital expenditure required to grow in the way we described, it is quite substantial. I think, Doug, our view on this is that it is a once-in-a-lifetime opportunity. The AI we are seeing is the biggest transformation of our lifetime. If we can achieve 39% growth on an annualized run rate of $184 billion, and then grow another 40% each year on that base for the next two years, eventually becoming a business over $1 trillion, we will invest capital to continue leading. We are already leaders in infrastructure and AI today. If this acceleration and opportunity were spread out over 25 years, could you spread capital expenditure over 25 years, would that be more convenient? Yes, perhaps it would be more convenient, but the world we live in is not like that, and the opportunity is not like that.
We can fund this plan. I think it will forever change Amazon's scale, forever change our revenue, and forever change our free cash flow. We will invest capital and pursue it relentlessly to continue being an important leader as we are today.
(VI) Q6 – Barclays Analyst Ross Sandler
Question: Okay, I want to dig deeper into the Bedrock question. The token marketplace on Bedrock seems to be one of the hottest services in the AI field currently. From our understanding, most of these workloads run on Trainium. As it becomes a more important part of the overall AWS story—as you just mentioned, Andy—how will this affect revenue growth, especially AWS margins, as you shift from selling bare metal to selling more inference as a service? Can you discuss this?
Andrew Jassy Response:
Yes. You are correct, currently most content on Bedrock runs on Trainium. An important reason we do this is that we really like the price-performance we achieve, which allows us to keep prices as low as possible for customers while also maintaining associated margins. If you think about margins in the cloud space, looking at the cost structures of different participants, there are a few key points to note. First is data centers, how much do they own or long-term lease, rather than using colo (colocation) or neocloud. Neocloud operating margins are about 3500 basis points worse than long-term leases.
Second, look at the hardware they run on. If you use hardware that everyone else uses, your cost structure will be comparable to others. If you have the ability to use different hardware, you have an advantage. So I think, looking at what we are doing—most of our capacity is owned or long-term leased, which enjoys better operating margins than those primarily adopting colo or neocloud, or a significant portion thereof. And we run on our own chips, which brings advantageous price-performance and is more cost-effective for us because it is proprietary—I think this gives us some structural cost advantages, which will be very beneficial to customers when we pass the savings on, and very beneficial to AWS overall economics.
(VII) Dave Fildes
Thank you everyone for attending today's conference call and asking questions. The replay will remain on our company's Investor Relations website for at least 3 months. Thank you for your attention to Amazon, and we look forward to speaking with you again next quarter.