Amazon's AWS Hits $150 Billion Run Rate With Fastest Growth in 15 Quarters, While Custom Chip Empire Quietly Crosses $50 Billion Shadow Revenue
Q1 2026 Earnings Call, April 29, 2026
Amazon delivered a quarter that surprised even seasoned AWS watchers, with two headline revelations that reframe the investment thesis: AWS growth is not only reaccelerating but doing so at a scale that has no modern precedent, and Amazon's custom silicon operation — long treated as a cost-savings footnote — has quietly become one of the largest data center chip businesses on the planet, with economics that could permanently reshape the company's margin structure.
AWS: Reacceleration Is Real, and the Numbers Are Striking
AWS posted $37.6 billion in revenue for the quarter, growing 28% year-over-year — an acceleration of 480 basis points from the prior period and the fastest growth rate in 15 quarters. CEO Andy Jassy was direct about the significance: "We haven't grown at this pace since we were about half the size." The business now runs at a $150 billion annualized revenue rate, and the sequential dollar addition — $2 billion quarter-over-quarter — was the largest Q4-to-Q1 increase in AWS history. AWS operating income came in at $14.2 billion.
What is driving the reacceleration is a combination of AI-native demand and a secondary wave of core infrastructure pull-through. CFO Brian Olsavsky noted that there is a strong empirical correlation between AI spending and core service growth: as customers move AI workloads into production, demand for compute, storage, databases and networking rises in tandem. Jassy elaborated that all the post-training, reinforcement learning, agentic actions and tool usage layered on top of frontier models are generating enormous CPU and core workload demand — something the market has underappreciated. The migration of enterprise workloads from on-premises to cloud is also quietly accelerating, partly because memory and storage supply chains are prioritizing the largest buyers, which are cloud providers, leaving smaller enterprises capacity-constrained on their own.
The AWS revenue backlog stood at $364 billion at the end of Q1. Crucially, Jassy noted this figure does not yet include the recently announced Anthropic deal worth over $100 billion. He also pushed back on the notion that backlog is concentrated: "There's reasonable breadth in that as well. It's not just one customer or two customers."
The Custom Silicon Revelation: A $50 Billion Business Hidden in Plain Sight
Perhaps the single most underappreciated disclosure of the call was Jassy's framing of Amazon's chip business. The company reported nearly 40% quarter-over-quarter growth in its chips segment, with an annual revenue run rate now exceeding $20 billion and growing at triple-digit percentages year-over-year. But the more striking figure came with an important clarification: "If our chips business was a stand-alone business and sold chips produced this year to AWS and other third parties as other leading chip companies do, our annual revenue run rate would be $50 billion."
Jassy went further, stating that Amazon's custom silicon operation is now, by its own assessment, one of the top three data center chip businesses in the world. The Trainium2 chip offers approximately 30% better price performance than comparable GPUs and is largely sold out. Trainium3, which began shipping at the start of 2026, offers 30% to 40% better price performance than Trainium2 and is nearly fully subscribed. Critically, much of Trainium4 — still roughly 18 months from broad availability — has already been reserved, suggesting demand visibility that extends well into 2028. Total revenue commitments for Trainium now stand at over $225 billion, including large multiyear, multi-gigawatt agreements with Anthropic and OpenAI, as well as commitments from Uber and others.
The margin implications are significant. Jassy stated explicitly: "At scale, we expect Trainium will save us tens of billions of dollars of CapEx each year and provide several hundred basis points of operating margin advantage versus relying on others' chips for inference." This is not aspirational language — it is a structural claim about the long-term cost architecture of the AWS business.
The company also disclosed that Graviton, its CPU chip, is being adopted by Meta to run CPU-intensive agentic AI workloads at scale, with Meta committing to tens of millions of Graviton cores. Jassy framed the broader point: "AI is commonly seen as a GPU story, but the rise of agentic workloads, real-time reasoning, code generation, reinforcement learning and multistep task orchestration is driving massive CPU demand as well." Graviton is used by 98% of the top 1,000 EC2 customers and delivers up to 40% better price performance than x86 alternatives.
Bedrock and the OpenAI Partnership: Strategic Depth, Not Just a Model Add
Amazon added OpenAI's models to Bedrock — GPT-5.4 went live the day before the call, with GPT-5.5 following shortly — but the more strategically meaningful announcement was the preview launch of Amazon Bedrock managed agents powered by OpenAI. Jassy described this as a stateful runtime environment that enables organizations to build generative AI applications and agents at production scale, and distinguished it sharply from today's stateless model APIs: "When you're building agents, you're building AI applications, you don't want to start anew every time you interact with the model. You want to store state, you want to store identity, you want to store what the conversation or the actions have been." He called stateful architecture "the future of how these agents are going to be built" and noted that no other provider currently offers it.
Bedrock itself is now used by over 125,000 customers, with nearly 80% of the Fortune 100 companies on the platform. Customer spend in Bedrock grew 170% quarter-over-quarter, and the platform processed more tokens in Q1 alone than in all prior years combined — a figure that illustrates how fast inference workloads are scaling. Amazon's AI revenue is growing triple digits year-over-year across AWS. Nearly all of Bedrock's inference workloads run on Trainium, which means efficiency gains in the chip directly translate into capacity expansion without proportional CapEx.
CapEx Trajectory: Elevated and Intentional
Cash CapEx reached $43.2 billion in Q1, primarily attributable to AWS and generative AI infrastructure. Jassy acknowledged the free cash flow headwind explicitly: "In times of very high growth like now, where the CapEx growth meaningfully outpaces the revenue growth, the early years free cash flow is challenged until these initial tranches of capacity are being monetized and revenue growth outpaces CapEx growth." He drew a direct parallel to the first major AWS growth wave and noted that AWS must lay out cash for land, power, buildings and hardware typically six to twenty-four months before billing customers, while data centers carry useful lives of 30-plus years and chips five to six years.
The company indicated high confidence in monetization, citing existing customer commitments covering a substantial portion of 2026 CapEx already authorized. Management also flagged that the faster AWS grows, the more short-term CapEx will be required — a trade-off they appear willing to accept given the scale of downstream revenue and free cash flow potential.
Amazon Leo: Larger Than the Market Appreciates, Still Unproven
Amazon Leo, the low-Earth orbit satellite broadband service, drew detailed commentary from Jassy for the first time with concrete revenue and operational parameters. With over 250 satellites in orbit, more than 20 launches planned for 2026 and over 30 in 2027, commercial service is on track for Q3 launch. Already-signed enterprise customers include Delta Airlines, JetBlue, AT&T, Vodafone, DIRECTV Latin America, NASA and others, with Delta committing at least half its fleet starting 2028.
Jassy made performance claims with precision: Leo will offer approximately 2x better downlink performance than existing alternatives and roughly 6x better uplink performance, at a cost advantage to customers. He described the long-term revenue opportunity as "a very large many billion-dollar revenue business" with financial characteristics "reminiscent of AWS — capital-intensive upfront where you're committing a lot of capital and cash in the early years for assets that you get to leverage over a long period of time."
The planned acquisition of Globalstar, which brings scarce global spectrum necessary for direct-to-device connectivity, and the agreement with Apple to power satellite services for iPhones and Apple Watches, extend Leo's addressable market meaningfully into consumer connectivity. CFO Olsavsky separately flagged a Q2 operating income headwind of approximately $1 billion related to Leo satellite manufacturing and launch costs, with capitalization of certain production costs expected to begin in Q4.
Retail: Record Unit Growth, Grocery Accelerating
On the retail side, unit growth of 15% year-over-year was the highest since the tail end of COVID lockdowns, while outbound shipping costs grew only 12% and fulfillment expense grew 9% on an FX-neutral basis — meaning the company is expanding volume faster than it is incurring incremental fulfillment costs. Jassy disclosed that Amazon is now the second-largest grocer in the United States, with over $150 billion in gross grocery sales in 2025. Perishable same-day delivery has grown more than 40x year-over-year and accounts for nine of the top ten most ordered items in same-day eligible markets. Whole Foods has over 550 locations with 100 more planned. Prime Day is moving to Q2 in most geographies in 2026, which partially inflates Q2 revenue guidance but creates a meaningful year-over-year comparison distortion given Prime Day was entirely in Q3 in 2025.
Advertising revenue came in at $17.2 billion, up 22% year-over-year, continuing to outpace the broader digital advertising market. Rufus, the agentic shopping assistant, showed monthly active users up 115% and engagement up nearly 400% year-over-year.
Q2 Guidance: Wide Range Reflects Real Uncertainty
Q2 net sales guidance of $194 billion to $199 billion implies approximately 15% to 18% year-over-year growth, with operating income guided between $20 billion and $24 billion — a notably wide $4 billion range. The range reflects Prime Day timing, stock-based compensation seasonality, the $1 billion Leo cost step-up, and fuel inflation affecting transportation costs that are only partially offset by a recently implemented FBA fuel and logistics surcharge. The company's highest-ever operating margin of 13.1% in Q1 is unlikely to be replicated in Q2 given these known headwinds.
The Internal AI Transformation: Concrete, Not Abstract
Jassy offered an unusually concrete internal case study on the productivity impact of agentic coding. An unnamed Amazon service underwent a full engine replacement — work that would historically require forty to fifty engineers and approximately a year — completed instead by five engineers in sixty-five days using agentic coding tools. "That is a very different world of operating," Jassy said. He extended the thesis broadly: "Every single one of our consumer businesses, every single one of our businesses in general" is being redesigned from scratch with AI-native assumptions, covering DevOps, customer service, research, analytics and sales operations.
The investor implication is straightforward but important: if Amazon can redeploy engineering capacity at this ratio across its operations, the medium-term operating leverage could be substantial, even before accounting for top-line AI revenue growth at AWS.
Amazon.com, Inc. Deep Dive
The Flywheel's Evolution: Business Model and Monetization
Amazon operates a highly diversified and vertically integrated business model that bridges retail, logistics, digital advertising, and enterprise cloud infrastructure. At its core, the company functions as a multi-sided platform. On the retail side, it generates revenue through first-party online sales, physical stores, and deeply entrenched third-party seller services. Third-party merchants now account for approximately 60 percent of gross merchandise volume, paying Amazon fulfillment fees, referral commissions, and advertising premiums. This ecosystem is stabilized by the Prime subscription model, which boasts over 260 million global members who pay recurring fees in exchange for accelerated shipping, digital media streaming, and exclusive discounts. The retail engine is characterized by immense scale and low operating margins, acting primarily as a high-velocity customer acquisition and retention funnel.
The true profit engines of the business, however, are found in its digital and enterprise segments. Amazon Web Services provides foundational cloud computing, storage, database, and machine learning infrastructure to millions of customers, ranging from early-stage startups to global enterprises and government agencies. By acting as the digital landlord for a significant portion of the internet, Amazon captures high-margin recurring revenue. Additionally, the company monetizes its massive retail traffic through a sophisticated digital advertising business. By integrating sponsored product placements and full-funnel video advertisements across its platform and digital properties, Amazon extracts a high yield from merchants seeking visibility. This structural duality allows Amazon to continuously reinvest the massive cash flows generated by its higher-margin divisions into capital-intensive physical and digital infrastructure, thereby reinforcing its competitive moats across all business lines. The success of this model is evidenced by the company generating 716.9 billion dollars in revenue in 2025 and 181.5 billion dollars in the first quarter of 2026.
Market Positioning and the Cloud Computing Battlefield
In the global cloud infrastructure market, Amazon Web Services remains the undisputed leader, maintaining approximately 31 percent market share as of early 2026. The division serves as the backbone of Amazon's profitability, boasting an annualized revenue run rate of 150 billion dollars and accelerating to a 28 percent year-over-year growth rate in the first quarter of 2026. Its customer base is highly diversified, capturing the majority of the top 500 venture-backed startups alongside massive enterprise workloads. However, the competitive landscape has intensified significantly during the generative artificial intelligence super-cycle. Microsoft Azure has closed the gap, capturing approximately 24 percent of the market. Microsoft leverages its deep integration with the ubiquitous Microsoft 365 enterprise software suite and an exclusive partnership with OpenAI to transition traditional infrastructure clients into sophisticated artificial intelligence contracts. Meanwhile, Google Cloud Platform holds around 12 percent of the market, differentiating itself through advanced data analytics, Kubernetes management, and proprietary tensor processing unit infrastructure.
The core threat to Amazon Web Services in this arena is the potential for workload defection as enterprises increasingly adopt multi-cloud strategies. Currently, around 87 percent of large organizations utilize multiple cloud providers, deliberately avoiding lock-in. While Microsoft captured significant early momentum in the artificial intelligence land grab by securing exclusive compute rights for OpenAI, Amazon has mounted a vigorous defense. The company's platform, Amazon Bedrock, positions itself as model-agnostic, allowing enterprise customers to deploy various foundation models, including those from Anthropic, Meta, and Cohere, securely within their existing virtual private clouds. The strategic objective is to prevent Microsoft and Google from utilizing artificial intelligence as a wedge to extract core compute and storage workloads. Early 2026 data indicates that this strategy is succeeding, with Amazon Web Services reporting its largest quarter-over-quarter revenue increase in history, signaling that the initial wave of experimental artificial intelligence spending is now translating into sustained production workloads on Amazon's infrastructure.
The Retail Fortress and the Cross-Border Insurgency
Domestically, Amazon's e-commerce dominance remains virtually unchallenged, capturing 37.6 percent of the United States online retail market. Its closest domestic competitor, Walmart, holds a 6.4 percent share, followed by platforms like Apple, eBay, and Target in the low single digits. Amazon's primary competitive advantage in retail is its unparalleled logistics and fulfillment network. By fully regionalizing its United States network, the company has localized inventory placement, significantly reducing the cost to serve while increasing delivery speeds. In 2025 alone, the company delivered over 8 billion items via same-day or next-day shipping, creating an operational moat that domestic competitors find structurally impossible to replicate at scale. This logistical efficiency is directly reflected in the North American segment's operating margins, which expanded to 9 percent by the end of 2025.
Despite this domestic stronghold, the global retail landscape has been disrupted by a new class of direct-from-factory Chinese platforms, commonly referred to as cross-border commerce or C-commerce. Entrants like Temu and Shein have bypassed traditional warehousing entirely, utilizing extreme supply chain flexibility, algorithm-driven manufacturing, and aggressive social media marketing to capture market share. In 2025, Temu matched Amazon's 24 percent share of the global cross-border e-commerce market and emerged as the most downloaded shopping app in the United States. These competitors initially thrived by exploiting de minimis trade loopholes, shipping ultra-low-cost parcels directly to consumers duty-free. However, as the regulatory environment tightened in late 2025 and 2026, these loopholes were largely closed, introducing tariffs and logistical friction that directly challenge the structural viability of the C-commerce model. In response to consumer demand for absolute value, Amazon launched Amazon Haul, a dedicated mobile storefront for items priced under 20 dollars. By utilizing direct-from-overseas shipping with two-week delivery timelines, Amazon is neutralizing the value proposition of Temu and Shein while protecting its core Prime delivery ecosystem from margin degradation.
Vertical Integration and the Silicon Moat
One of the most consequential shifts in the technology sector is the realization that artificial intelligence workloads, particularly generative models, require an unprecedented volume of computing power, heavily concentrating supplier power in the hands of graphics processing unit manufacturers like Nvidia. Recognizing this supply chain vulnerability and the severe margin implications of generic hardware dependency, Amazon has aggressively pursued vertical integration at the semiconductor level. The company's custom silicon portfolio, specifically the Graviton central processing units, and the Trainium and Inferentia artificial intelligence accelerators, has evolved into a formidable competitive advantage. By early 2026, the custom silicon division achieved a 10 billion dollar annualized revenue run rate. Graviton chips now handle the majority of new elastic compute cloud workloads, delivering up to 40 percent better price-to-performance ratios than leading x86 processors, and are utilized by over 90 percent of Amazon's top enterprise customers.
As the artificial intelligence market transitions from the capital-intensive training phase to the continuous inference phase, cost-to-serve becomes the defining metric. Generic graphics processing units are highly versatile but inherently expensive to operate at scale for specific inference tasks. Amazon's Inferentia and the newly deployed Trainium2 architectures are purpose-built to optimize these precise workloads, drastically lowering the internal cost of compute. By controlling the hardware stack, Amazon can offer aggressive pricing to artificial intelligence developers without compressing its own operating margins. Furthermore, this vertical integration acts as an effective defense against emerging artificial intelligence chip startups like Cerebras and Groq. While these new entrants are building highly efficient inference architectures, Amazon's ability to offer its own low-cost, tightly integrated silicon within the world's largest cloud ecosystem limits the penetration potential of standalone hardware challengers.
The Margin Engines: Advertising and Subscriptions
Amazon's digital advertising business has quietly matured into a global powerhouse, generating over 68 billion dollars in 2025 and establishing the company as the third-largest digital advertising platform worldwide, commanding an 11 percent market share. Unlike Google or Meta, which rely heavily on probabilistic targeting across the open web, Amazon benefits from deterministic, closed-loop attribution. Consumers on Amazon exhibit high commercial intent, allowing the company to charge premium cost-per-click rates, which averaged 1.12 dollars in 2025. Advertisers are willing to absorb these rising costs because the return on ad spend is instantly measurable and directly tied to point-of-sale conversions. This dynamic creates a high-margin revenue stream that essentially subsidizes the capital expenditures required for the retail fulfillment network.
The company is now aggressively expanding this high-margin revenue through a full-funnel advertising strategy. The introduction of ad-supported tiers on Prime Video brought in an estimated 12 billion dollars in incremental revenue in 2025. By blending top-of-funnel brand awareness campaigns on streaming television with bottom-of-funnel sponsored product listings, Amazon is capturing media dollars that historically went to linear television or traditional search platforms. This expansion is further augmented by the integration of generative artificial intelligence tools that allow advertisers to instantly create campaign creatives and optimize bidding strategies based on real-time retail signals. The compounding effect of subscription revenue from 260 million Prime members and high-margin advertising real estate ensures that the retail side of the business remains structurally profitable even during periods of heavy reinvestment.
Next-Generation Growth Drivers
Beyond its core operations, Amazon is cultivating several new technologies positioned to drive future growth. In the enterprise software sector, the company has launched Amazon Q, a generative artificial intelligence assistant designed to integrate deeply into corporate workflows. Unlike broad consumer models, Amazon Q connects directly to secure enterprise data silos, including internal wikis, Salesforce, and Microsoft 365, allowing employees to query proprietary data while strictly adhering to corporate access controls. Furthermore, Amazon Q Developer functions as a sophisticated coding companion, assisting software engineers with code generation, legacy application modernization, and security patching. This represents a direct assault on Microsoft's Copilot ecosystem, aiming to capture the highly lucrative per-seat enterprise software licensing market.
In the physical infrastructure space, Project Kuiper represents a massive, long-term capital bet on low Earth orbit satellite broadband. Having deployed 180 satellites by the end of 2025, Amazon is initiating commercial operations in 2026. Rather than focusing solely on direct-to-consumer rural broadband, the company has secured significant enterprise and telecommunications partnerships with entities like AT&T and JetBlue. Project Kuiper serves a dual purpose: it creates a new recurring revenue stream in global connectivity and acts as an extension of the Amazon Web Services edge computing network, allowing multi-national enterprises to process data in remote geographic locations and seamlessly pipe it back into Amazon's cloud infrastructure.
Management Track Record: From Austerity to Aggressive Expansion
Under the leadership of Chief Executive Officer Andy Jassy, Amazon has demonstrated an exceptional ability to quickly alter its operational posture in response to macroeconomic realities. Inheriting an overbuilt logistics network post-pandemic, management executed a ruthless optimization strategy. By overhauling the fulfillment architecture from a national model to a highly regionalized system, they simultaneously stripped out billions in fixed costs and improved delivery speeds. This era of operational austerity culminated in the first quarter of 2026, when the company reported a record-breaking overall operating margin of 13.1 percent and an operating income of 23.9 billion dollars.
Having stabilized the retail margins, management is now executing a massive pivot back toward aggressive expansion, guided by the conviction that generative artificial intelligence represents a generational platform shift. The company has guided for approximately 200 billion dollars in capital expenditures for 2026, the vast majority of which is allocated to securing real estate, power, and compute capacity for Amazon Web Services. This astronomical spending level has understandably drawn scrutiny regarding its impact on near-term free cash flow. However, management has transparently communicated that the bottleneck in the artificial intelligence market is currently data center capacity, not enterprise demand. The track record of the leadership team suggests a proven capability to digest massive capital cycles and ultimately yield outsized recurring cash flows once the underlying infrastructure is monetized.
The Scorecard
Amazon has successfully transitioned from a low-margin retail aggregator into an extraordinarily profitable infrastructure utility that underpins both global commerce and enterprise computing. The regionalization of its fulfillment network has fortified the retail division against domestic challengers, while the integration of its 68 billion dollar advertising segment ensures structural profitability even in a mature e-commerce market. In the cloud computing domain, the initial momentum lost to competitors during the early days of the generative artificial intelligence boom has been decisively reclaimed. By leveraging a model-agnostic software approach alongside deep vertical integration in custom silicon like Graviton and Trainium, Amazon is structurally insulated against the margin degradation associated with external hardware dependency. The first quarter of 2026 demonstrated the sheer operating leverage of this model, resulting in the highest operating margins in the company's history.
The primary risks moving forward revolve around capital intensity and the shifting regulatory landscape. The decision to deploy 200 billion dollars in capital expenditures in 2026 represents a massive strain on free cash flow, requiring an assumption that downstream enterprise artificial intelligence monetization will remain robust for years to come. Furthermore, while the cross-border threat from entrants like Temu and Shein has been blunted by regulatory tariffs, the global retail market remains highly sensitive to aggressive pricing strategies. Ultimately, Amazon's unparalleled scale, integrated logistics moat, and control over its semiconductor destiny provide it with the structural resilience needed to absorb these competitive shocks. The company is uniquely positioned to capture the economic surplus of the artificial intelligence era, not merely as a software provider, but as the foundational physical and digital infrastructure of the modern economy.