Operating income has grown sixfold since 2022. The market still thinks it is buying a retailer.
Amazon's operating income in 2025 was $80 billion. In 2022 it was $12.2 billion. That is a 555% increase in three years, on revenue that grew 39%.
That kind of operating leverage does not happen by accident. It happens when two high-margin businesses, cloud computing and digital advertising, grow fast enough to overwhelm a historically low-margin retail operation that still anchors most investors' mental model of the company.
The market has not fully priced this shift. At 15.4x EV/EBITDA and 3.58x price-to-sales, Amazon is trading at multiples that would look reasonable for a mature consumer retailer. The business that actually generates the profits is something else entirely. That gap is the investment thesis.
Amazon is three businesses sharing one balance sheet. The first is the retail and logistics operation: the marketplace, first-party inventory, and the fulfillment network. It is enormous, growing, and generates thin margins. It is also the part of the business that most investors have spent 20 years watching.
The second is Amazon Web Services. AWS is the world's largest cloud infrastructure provider, with an estimated operating margin in the 35 to 40 percent range. It generates recurring revenue from enterprise customers who have built critical systems on AWS infrastructure and face significant switching costs. That is a structurally different business from selling books.
The third is Advertising. Amazon's advertising business crossed $50 billion in annual revenue in 2024, growing at rates that outpace Google and Meta in comparable segments. This is a very high-margin business. Advertisers pay Amazon to reach consumers at the exact moment of purchase intent, which is a placement no other platform can replicate at scale.
These three businesses are consolidated into a single P and L. The retail segment obscures what the other two are producing. Understanding Amazon in 2026 requires disaggregating them.
TickerXray Report
Get the complete Amazon report with all 12 quantitative models, AI-generated investment thesis, and real-time data.
Amazon's top line grew from $469.8 billion in 2021 to $716.9 billion in 2025, a 53% cumulative increase. That is solid but unremarkable for a company of this scale.
The operating income trajectory is where it gets interesting. Amazon posted $24.9 billion in operating income in 2021, then compressed to $12.2 billion in 2022 as the post-pandemic retail normalization hit and logistics costs surged. That compression spooked investors and drove the stock's 2022 selloff. What came next was the setup investors missed.
Operating income recovered to $36.9 billion in 2023, accelerated to $68.6 billion in 2024, and reached $80 billion in 2025. Operating margin went from 2.4% in 2022 to 11.2% in 2025. Gross margin expanded from 42.0% in 2021 to 50.3% in 2025, reflecting the structural shift in revenue mix toward higher-margin services.
Net income went from negative $2.7 billion in 2022 to $77.7 billion in 2025. EBITDA reached $165.3 billion. These are not the numbers of a retail company. They are the numbers of a technology conglomerate that happens to ship packages.
Amazon's free cash flow compressed dramatically in 2025, falling to $7.7 billion from $32.9 billion the prior year. This number has attracted bearish commentary. The context matters enormously.
Capex in 2025 was $131.8 billion, up from $83 billion in 2024 and $52.7 billion in 2023. This is the most aggressive capital investment program in the company's history. The target is AI infrastructure: data centers, custom silicon, network capacity for AWS and the emerging AI services stack. Operating cash flow, which strips out the timing distortions of working capital, reached $139.5 billion in 2025, the highest in company history.
The distinction matters. When a capital-intensive business generates $139.5 billion in operating cash flow and chooses to invest $131.8 billion back into future capacity, the resulting $7.7 billion in free cash flow is a policy decision, not an earnings problem. The company is investing because the returns on incremental AWS and AI infrastructure are high enough to justify it.
The parallel to look at is Amazon in 2013 to 2016, when the company was cash flow negative while building the infrastructure that would eventually generate the 2025 numbers. Investors who confused investment intensity with financial weakness at that point paid a significant opportunity cost.
For context, the balance sheet remains healthy: $86.8 billion in cash, $65.6 billion in long-term debt, and $411.1 billion in total equity. The company is investing from a position of strength.
AWS has been the primary driver of Amazon's margin expansion. Cloud infrastructure demand has grown faster than consensus expected, and AWS has maintained pricing discipline despite competition from Azure and Google Cloud. Enterprise AI workloads are accelerating this further. Customers adopting large language models and inference pipelines at scale are doing so primarily on cloud infrastructure, and AWS is the largest provider of that infrastructure.
On April 11, Andy Jassy signaled to investors that AWS momentum is continuing into 2026, describing capacity expansion plans and AI service demand as the primary growth vectors for the year ahead. The specifics of that announcement, including expanded inference capacity and new agent frameworks for enterprise customers, are consistent with a business that is growing into its capex rather than wasting it.
The advertising business deserves more attention than it typically receives. Amazon Ads benefits from something neither Google Search nor Meta can replicate: purchase-intent data at the exact moment of transaction. When a consumer searches for a product on Amazon, the intent signal is unambiguous. Advertisers pay a meaningful premium for that placement. This business is growing at rates that outpace the consolidated company, carries margins comparable to the best pure-play ad businesses, and is still early in its monetization of the audience it has built.
These two segments are why gross margin has expanded from 42% to 50.3% in four years. The retail business has not become dramatically more efficient. The mix has shifted.
Amazon trades at 15.4x EV/EBITDA. That multiple would be appropriate for a diversified industrial conglomerate or a mature retailer with steady but slow growth. It is not the multiple typically assigned to a business generating $80 billion in operating income and growing that figure sixfold in three years.
For comparison, Microsoft trades at roughly 20 to 25x EV/EBITDA. Google trades around 15 to 18x. The companies that most resemble the AWS and Advertising profit profile trade at premiums to Amazon's current multiple. The persistent discount reflects two things: the historical retail framing that still dominates analyst mental models, and the near-term free cash flow compression from the capex ramp, which makes the stock look expensive on traditional FCF yield metrics.
The trailing P/E is 33.2x, which on its face looks elevated. But net income of $77.7 billion on a market cap of approximately $2.56 trillion means the earnings yield is 3.0%. On a business with demonstrated operating leverage and a high-return reinvestment cycle, that is not an expensive entry point.
Analyst consensus reflects this view. Of 70 analysts covering the stock, 47 have a strong buy, 19 have a buy, and 4 have a hold. Zero have sell or strong sell ratings. The average price target of $281.27 implies 20 to 25% upside from current levels.
Amazon's retail network is self-reinforcing in ways that are structurally difficult to replicate. More sellers attract more buyers. More buyers attract more sellers. The logistics network gets more efficient as volume increases. Prime membership creates behavioral lock-in that raises switching costs without requiring contractual obligations. This flywheel has been documented extensively and remains intact.
What has changed is the overlay of AI capabilities across the business. Amazon has developed its own custom silicon, the Trainium and Inferentia chips, to reduce reliance on third-party GPU suppliers and lower the unit economics of running AI workloads on AWS. The April 12 reports of Amazon's Kuiper satellite broadband ambitions suggest the company is also investing in last-mile connectivity infrastructure, which would extend the addressable market for AWS edge computing while adding a potential competitor to Starlink in enterprise and remote connectivity markets.
Agentic AI applications, which received significant attention in April 2026 coverage of the sector, are particularly well-suited to cloud infrastructure at scale. The companies building those applications need reliable, low-latency compute and storage. AWS is positioned to capture a disproportionate share of that workload.
The capex commitment is the most concrete near-term risk. Amazon has committed to $131.8 billion in capital expenditures in 2025, and the 2026 figure is expected to remain elevated. If AI workload demand softens, or if the return on incremental AWS capacity comes in below expectations, the company will have overbuilt infrastructure at significant cost. This is the same risk Microsoft, Google, and Meta are all running simultaneously. The question is not whether any of them overbuilt, but by how much and for how long.
Antitrust exposure is the second risk, and it has become more specific over the past 18 months. Regulatory scrutiny of Amazon's dual role as both a marketplace operator and a seller on that marketplace has intensified in the US and Europe. Potential remedies could include structural separation of the marketplace and first-party retail operations. This would not destroy the AWS or advertising businesses, but it would alter the retail competitive dynamics in ways that are difficult to model.
The third risk is AWS margin compression. Cloud pricing has been deflationary over time as providers compete for large enterprise contracts. If Azure or Google Cloud accelerate market share gains through aggressive pricing or superior AI tooling, AWS pricing power could erode. The current 10.8% to 11.2% consolidated operating margin assumes AWS continues to expand or hold margins. A meaningful compression in AWS pricing would flow directly to the consolidated income statement.
Amazon in 2026 is not the company it was in 2019. Revenue has grown, but the operating income has grown at a rate that reveals a fundamentally different business beneath the logistics operation that consumes most of the media coverage.
At 15.4x EV/EBITDA, the market is pricing the blended business roughly in line with slower-growing industrial conglomerates. The component businesses, AWS and Advertising, would each command a premium multiple if they traded independently. The retail operation acts as a discount anchor.
The capex-driven FCF compression in 2025 is real but analytically distinct from an earnings problem. The company is investing $131.8 billion because the return on that investment, primarily in AI infrastructure and cloud capacity, is expected to exceed the cost of capital by a significant margin. The operating cash flow of $139.5 billion tells the underlying story.
This is a business with a 50.3% gross margin, consistent earnings beats across six consecutive quarters, and an operating income trajectory that has grown sixfold in three years. The multiple assigned to it reflects the old story. The data supports a different one.
Full forensic analysis of Amazon
+ 6 more models included
150,000+ stocks covered
Global coverage across 60+ exchanges. Every report includes all 12 quantitative models and AI analysis.
View plansEvery report runs 12 quantitative models and generates an AI investment thesis. From Piotroski scores to manipulation detection -- get the full picture in seconds.
12 forensic models
Piotroski, Altman, Beneish, DuPont & more
AI investment thesis
Synthesized outlook on every stock
Manipulation detection
Spot red flags before they hit the news
150,000+ tickers
Global coverage across 60+ exchanges
Expected return
Forward return projections for every stock
Real-time data
Live prices, insider trades, news sentiment
Free accounts get 1 report per month. Pro gets unlimited.
From 5.3% to 11.2% operating margin in four years. Now the company is spending $131.8 billion in a single year. This is either the best capital allocation decision in tech history or a very expensive bet on demand that may not materialise.
The market still prices Amazon as a retailer. The financials say otherwise.
Andy Jassy's shareholder letter explicitly embraces ROIC and capital efficiency for the first time. With operating margins at 10.5% and $77.7B in net income, the margin story we flagged is accelerating.