Amazon is a composite platform that grafts low-margin retail traffic onto advertising and AWS, where AWS makes up 18% of revenue yet contributes 57% of operating profit. Rating: Watch—a great company, but not at a great price.
The tension is crisp: at the operating level it is a top-tier cash machine, with 2025 operating cash flow near $140 billion; yet at the shareholder-distributable level it nearly drops to zero, as AI capex squeezed TTM free cash flow down to just $1.2 billion. The report's mid DCF value across three tiers is about $174, while the current price of $264.86 already sits against the upper bound of the optimistic scenario. What is genuinely fragile is the quality of growth, not growth itself—the market is betting that incremental AI capital can replicate AWS's historical returns, and the moment results are merely "good rather than stunning," the valuation multiple has to compress.
The line to watch most is AWS share being chased by Azure/Google and its margin falling below 30%. The report sets an ideal buy of $120-155 and an acceptable range of $155-220; if the bull thesis is damaged, a medium-to-long-term drawdown of 35%-55% is not unimaginable. First grant that it is excellent, then grant that it is not cheap right now.
Labeling convention: Below I try to separate Fact, Assumption, Inference, and Opinion.
Fact: drawn from Amazon's annual reports, quarterly filings, proxy statements, official investor relations pages, SEC filings, and public sources such as the U.S. Treasury or the U.S. Census Bureau.
Assumption: the growth rates, discount rates, maintenance capex ratios, and so on used in the valuation.
Inference: business and valuation judgments derived from the facts.
Opinion: the investment conclusions formed on the basis of the facts and inferences above.
Conclusion First
Investment Rating: Watch
Core Judgment Amazon is a business I can understand, though hardly a "simple" one: a compounding enterprise whose base layer is its traffic and fulfillment network, whose middle layer is third-party sellers and Prime membership, and whose top layer is advertising and AWS, with profit heavily concentrated in the high-margin services businesses. In 2025 the company posted revenue of 716.9 billion dollars, operating profit of 80.0 billion dollars, and operating cash flow of 139.5 billion dollars, a business quality clearly stronger than most retailers. But 2025 net capex rose sharply to 128.3 billion dollars, dragging free cash flow down to roughly 11.2 billion dollars, and in Q1 2026 the company also stated explicitly that TTM free cash flow was only 1.2 billion dollars, mainly because of a marked increase in AI-related capital spending. In other words, Amazon is still a top-tier enterprise, but at this stage it looks more like a "high-quality, high-reinvestment, high-valuation" watch-list candidate than an obvious margin-of-safety buy.
Is there a margin of safety at the current price: not obvious
As of the tool snapshot on May 19, 2026, AMZN traded at roughly 264.86 dollars, implying a market cap of about 2.88 trillion dollars and a trailing P/E of roughly 31.7x. Against the conservative, neutral, and optimistic intrinsic-value bands I lay out later, the current price sits roughly near the upper edge of my optimistic case, yet well above the conservative and neutral valuation ranges, so I do not see a sufficiently clear margin of safety.
Suitable Investor Profile This better suits long-term growth-oriented value investors willing to track business quality over time, accept valuation swings, and understand that "free cash flow over the short to medium term will be swallowed by AI infrastructure"; it is less suited to ordinary value investors who only want to buy "undervalued assets" or who rely on the current free-cash-flow yield.
Greatest Uncertainty There are three key uncertainties:
Whether the enormous capital spending of the AI era can ultimately earn returns comparable to those of AWS historically.
Whether AWS can keep its share and margins resilient as Azure and Google Cloud close the gap.
Whether regulatory and antitrust risk reshapes the economics of Marketplace, advertising, Prime, or the Featured Offer.
The Conclusion in One Sentence If the stock market were to close for the next five years, I would be willing to own the Amazon business; but if I had to decide today whether to buy the whole company outright at the current price, I would choose to keep watching and wait for a better return-to-risk ratio. This is a textbook case of "a good company, but not at an obviously cheap price right now."
Understanding the Business and the Industry Landscape
How exactly does this company make money
Fact: For disclosure purposes Amazon splits into three segments, North America, International, and AWS; its customer set spans consumers, sellers, developers, enterprises, content creators, advertisers, and employees. Its main revenue types include online retail, third-party seller services, subscription services, advertising services, and AWS. The company discloses clearly that third-party seller services are charged through commissions and fulfillment/delivery fees; advertising is charged per click or impression; subscriptions such as Prime are recognized over the subscription period; and AWS is recognized mainly by usage or contract term.
Fact: In 2025 Amazon's revenue groupings were: online stores 269.29 billion dollars, physical stores 22.56 billion dollars, third-party seller services 172.16 billion dollars, advertising 68.64 billion dollars, subscriptions 49.62 billion dollars, AWS 128.73 billion dollars, and other 5.94 billion dollars, for a total of 716.92 billion dollars. 2025 segment operating profit was: North America 29.62 billion dollars, International 4.75 billion dollars, and AWS 45.61 billion dollars, for a total of 79.98 billion dollars. From this one can infer that AWS accounts for only about 18% of revenue yet contributes roughly 57% of operating profit; the North America retail ecosystem contributes the bulk of the rest, while the international business is far thinner on profit.
Inference: So the real commercial nature of Amazon is not that of a "single retailer" but of a multi-layered flywheel.
The first layer is the consumer traffic and fulfillment network, which delivers scale and data.
The second layer is seller services and Prime membership, which strengthen repeat purchasing and platform stickiness.
The third layer is advertising and AWS, which monetize that traffic and technical capability and output high profit. What a long-term business owner should watch is not online retail GMV itself, but seller penetration, advertising monetization, AWS growth, and the return on capital spending.
Is the revenue recurring, stable, and predictable
Fact: Amazon's deferred revenue comes mainly from AWS prepayments and Prime membership; at year-end 2025 deferred revenue was 25 billion dollars. In addition, the company discloses that future unsatisfied performance obligations with an original term of more than one year are mainly AWS-related, at roughly 244 billion dollars at year-end 2025, with a weighted-average remaining term of 4.1 years. This means AWS has clear contractual visibility, while Prime also provides prepaid stickiness.
Inference: Amazon's revenue stability is "layered stability."
1P retail is the thinnest and most exposed to the consumer cycle and competition.
3P services, advertising, and Prime are more stable.
AWS has the strongest customer stickiness and contractual visibility. So this is not a business with "constant revenue," but it is a business with steadily rising recurring revenue.
Cost Structure and Dependencies
Fact: Amazon discloses clearly that its variable costs include the cost of goods and content, payment processing, picking and packing, shipping, customer service, and the cost of running AWS; the company also stresses that its goal is to turn inventory quickly and to benefit from an operating cycle in which consumer collections precede payments to suppliers. In the 2025 cost structure, cost of sales was 49.7% of net sales, fulfillment 15.2%, and technology and infrastructure 15.1%. 2025 shipping costs were 102.7 billion dollars.
Inference: This business does not depend on any single customer, but it does depend on several key resources:
A vast warehousing and delivery network.
Data centers, compute, energy, and the chip supply chain.
The third-party seller ecosystem.
A sustained high level of engineering and operational talent. It is not a business "transparent enough to read at a glance," because retail, advertising, cloud, content, logistics, hardware, and healthcare are all mixed together, but the core money-making logic is understandable.
If the stock market closed for five years, would I be willing to hold
My answer is: yes, provided the entry price is reasonable. Amazon is not a company that pumps short-term profit through financial engineering; its real value comes from a long-built network, a rising share of services revenue, and the high-return flywheel of AWS and advertising. It is just that the current price already prices in a fair amount of expectation for future AI investment returns.
Business Comprehensibility Score: 4/5
It is not a simple business, but its core flywheel is understandable: retail traffic pool + seller services + advertising + cloud computing. The complexity comes from having too many businesses, very large capital spending, and a fair amount of accounting noise.
Industry and Competitive Landscape
Fact: The U.S. Census Bureau reports that in Q1 2026 U.S. online retail sales were about 302.3 billion dollars, up 9.7% year over year and accounting for 16.8% of total retail sales; Digital Commerce 360 estimates 2025 U.S. e-commerce sales at about 1.234 trillion dollars, up 5.4% year over year. This shows that e-commerce is still growing, but is no longer in the early, linear "wild expansion" phase.
Fact: A Synergy Research release shows that in Q3 2025 the global cloud infrastructure services market was about 106.9 billion dollars, with AWS, Microsoft, and Google holding roughly 29%, 20%, and 13% share respectively; a publicly cited Synergy reading for Q1 2026 shows AWS still in the lead with about 28% share, Azure about 21%, and Google Cloud about 14%.
Inference: Amazon is really not in one industry but in three industries stacked together:
E-commerce and retail: large and stable, but inherently low-margin and fiercely competitive.
Cloud infrastructure: high growth, high capex, high technical barriers.
Retail media advertising: fast growth, high margins, benefiting from closed-loop platform data. So Amazon is not "a single good company in a good industry"; it is a composite that grafts relatively low-return retail demand onto the higher-return advertising and AWS businesses. That structure is itself an important source of the company's quality.
Main Competitors and Industry Position
The competitors Amazon itself lists span physical retail, e-commerce, media, search and social entry points, cloud services, fulfillment logistics, digital advertising, and healthcare services. Its most important real rivals are: on the retail side, Walmart, Costco, and various platforms; on the cloud side, Microsoft Azure and Google Cloud; and on the advertising side, Google and Meta. In terms of scale and integrated capability, Amazon remains one of the most important platform companies in the world, especially at the intersection of e-commerce infrastructure, retail media, and the public cloud, where it has almost no fully substitutable rival.
Industry Attractiveness Score: 4/5
Demand is stable over the long run and platform-type winners capture the scale dividend, but technology turnover, regulation, and capex intensity are all high. The industry is attractive, but by no means easy.
Moat and Management
Moat Analysis
I break Amazon's moat into ten elements:
| Moat Element | Assessment | Main Basis |
|---|---|---|
| Brand advantage | Strong | Amazon, Prime, and AWS have become default options for consumers and enterprises. |
| Cost advantage | Strong | Scale procurement, inventory turnover, fulfillment density, and platform-style seller economics deliver unit-cost advantages. |
| Scale advantage | Very strong | 2025 revenue of 716.9 billion dollars, net property and equipment of 357.0 billion dollars, and 2025 net capex of 128.3 billion dollars. |
| Network effects | Moderately strong | More consumers attract more sellers; more sellers raise the value of advertising inventory and fulfillment. |
| Switching costs | Moderately strong | AWS carries clear migration friction in cloud architecture, data, and toolchains. |
| Channel advantage | Very strong | Self-built logistics, warehouse network, membership system, device entry points, and traffic entry points work in concert. |
| Patent/license/regulatory barriers | Moderate | Not the core barrier, but cloud, healthcare, payments, and other businesses carry compliance thresholds. |
| Data advantage | Very strong | Transaction, click, search, delivery, advertising, and cloud-operations data form a closed loop. |
| Culture/operating capability | Very strong | A long-standing emphasis on customer obsession, long-term thinking, and operational excellence. |
| Capital allocation capability | Moderately strong | Historically built AWS and the fulfillment network through internal investment, but future AI returns remain to be proven. |
These assessments rest on Amazon's disclosures about its customer set, revenue model, and operating and cost structure, together with its 2025 revenue mix, profit mix, and scale of capital spending.
Is this moat widening, stable, or narrowing
My judgment: overall "stable to widening," but with internal divergence.
The data closed loop across retail logistics, membership, third-party sellers, and advertising is still widening the moat.
AWS is still very strong, but in the AI cycle Azure and Google are catching up faster, and AWS's relative edge is less comfortable than before. So one cannot simply say Amazon's moat is widening across the board. The more accurate statement is: the platform and fulfillment moat is widening, the cloud relative-share moat is under pressure, but the overall moat remains very deep.
How long and how much capital would a competitor need to replicate it
Replicating any single subsystem of Amazon is hard enough; replicating the combination of "retail traffic + membership + seller network + advertising + global cloud + fulfillment infrastructure" typically requires years and sustained investment on the order of tens of billions of dollars or more. In 2025 alone Amazon's net property and equipment reached 357.0 billion dollars, with net capex of about 128.3 billion dollars that year. This cannot be quickly replicated through financing alone; it depends far more on long-term operational accumulation.
Can it raise prices in an inflationary environment, and stay profitable in a downturn
Amazon has partial pricing power, but it is concentrated mainly in AWS, advertising, and subscriptions rather than in 1P retail. On the retail side it more often passes efficiency gains back to customers, building a "low-price mindset," so it is not a typical consumer-goods price-raiser. At the company level it still has strong staying power in a downturn: in 2022, even with negative net income, operating cash flow was still 46.75 billion dollars; 2025 and Q1 2026 show that AWS and advertising profit can materially cushion retail swings.
Were the high margins of the past a structural advantage or a cyclical windfall
My judgment is: mainly a structural advantage, with some cyclical and accounting noise mixed in. The structural part comes from the rising share of services revenue, especially AWS, advertising, subscriptions, and seller services; the noise part comes from fair-value changes on equity investments, adjustments to depreciation lives, and FTC and tax settlements. The large jump in net income in Q1 2026 was largely driven by a 16.8 billion dollar pre-tax gain from the Anthropic investment, which should not be treated simply as sustainable operating profit.
Moat Strength Score: 4.5/5
This is a world-class moat, but not one that is "never under threat." The greatest test comes from the capital war and the share war of the AI cloud era.
Management and Capital Allocation
Fact: Jeff Bezos remains Executive Chairman, holding about 950 million shares, or 8.8%; Andy Jassy holds about 2.31 million shares, under 1%. The proxy shows that executive compensation is mainly long-term RSUs, usually vesting over five years or more, with no annual cash bonus; after Jassy received the large CEO-related RSU grant in 2021, the committee made no new equity grant to him through 2025. The company did no buybacks in 2023-2025 and pays no dividend, with 6.1 billion dollars of buyback authorization remaining at year-end 2025.
Inference:
Honesty: high overall. The company clearly discloses which parts of current profit come from FTC settlements, layoffs, tax disputes, or investment gains, and states directly that the decline in free cash flow stems mainly from AI investment.
Shareholder alignment: Bezos is highly aligned with shareholders; Jassy's personal stake is not very large, but his compensation structure is long-term and equity-based, still more long-term than that of most professional managers.
Capital allocation: historically very successful, with the most classic examples being the long-term investments in AWS and the fulfillment network; but over the next three to five years, the most important capital-allocation test will be the return on AI infrastructure investment.
Supplementary Judgment on M&A and Governance
The focus of Amazon's capital use in recent years has clearly been internal reinvestment rather than returning capital to shareholders through large buybacks or dividends. As for whether M&A creates value, the public segment disclosure is not enough for me to draw a high-confidence conclusion on One Medical, MGM, and the like, so here I can only offer a cautious judgment: M&A is not the core of the current investment thesis, nor a reason for me to pay a high valuation. One governance point to watch is that the company has transactions with Bezos-affiliated entities; for example, the proxy discloses a 2.7 billion dollar satellite-launch agreement with Blue Origin in 2022, though the Board also stresses that such matters are overseen by an independent directors' committee. This is not disqualifying, but it is a governance point that warrants ongoing attention.
Management and Capital Allocation Score: 4/5
A high score historically, with the future yet to be tested. The biggest positive is the long-term orientation and capacity to reinvest; the biggest deduction is the uncertainty around the current return on AI investment.
Financial Quality
First, the core financial performance. All figures in the table below are in billions of dollars, and free cash flow is calculated as far as possible on Amazon's usual basis of "operating cash flow less net capex."
| Year | Revenue | Operating profit | Net income | Operating cash flow | Free cash flow | Net capex | Operating margin | Net margin |
|---|---|---|---|---|---|---|---|---|
| 2017 | 177.9 | 4.1 | 3.0 | 18.4 | 8.3 | 10.1 | 2.3% | 1.7% |
| 2018 | 232.9 | 12.4 | 10.1 | 30.7 | 19.4 | 11.3 | 5.3% | 4.3% |
| 2019 | 280.5 | 14.5 | 11.6 | 38.5 | 25.8 | 12.7 | 5.2% | 4.1% |
| 2020 | 386.1 | 22.9 | 21.3 | 66.1 | 31.0 | 35.0 | 5.9% | 5.5% |
| 2021 | 469.8 | 24.9 | 33.4 | 46.3 | -9.1 | 55.4 | 5.3% | 7.1% |
| 2022 | 514.0 | 12.2 | -2.7 | 46.8 | -11.6 | 58.3 | 2.4% | -0.5% |
| 2023 | 574.8 | 36.9 | 30.4 | 84.9 | 36.8 | 48.1 | 6.4% | 5.3% |
| 2024 | 638.0 | 68.6 | 59.2 | 115.9 | 38.2 | 77.7 | 10.8% | 9.3% |
| 2025 | 716.9 | 80.0 | 77.7 | 139.5 | 11.2 | 128.3 | 11.2% | 10.8% |
In the table, 2017-2019 data come from the 2018 and 2019 10-Ks; 2020-2022 from the 2021 and 2022 10-Ks; and 2023-2025 from the 2023 and 2025 10-Ks and annual-report figures.
How to read this table
Fact: From 2017 to 2025 revenue grew from 177.9 billion dollars to 716.9 billion dollars, an eight-year compound growth rate of about 19%; looking only at 2020-2025, the compound rate is about 13%. The operating margin improved from 2.3% in 2017 to 11.2% in 2025, which is not a typical retail trajectory but the result of a rising share of high-margin services revenue.
Fact: In 2025 cost of sales was 49.7% of net sales, implying a gross margin of about 50.3%; the corresponding 2024 figure was about 48.9%. Looking back, cost of sales was 58.0% in 2021, 56.2% in 2022, and 60.4% in 2020. In other words, Amazon's economic structure has been shifting over the past few years from "merchandise-retail driven" toward "services-and-platform driven."
Fact: But free cash flow is far more volatile than net income. In 2025 operating cash flow reached 139.5 billion dollars, but net capex rose to 128.3 billion dollars, leaving free cash flow at only about 11.2 billion dollars; in Q1 2026 the company further disclosed that TTM operating cash flow had risen to 148.5 billion dollars, but TTM free cash flow had fallen to 1.2 billion dollars, mainly because net capex rose 59.3 billion dollars year over year and "primarily reflects AI investment."
Inference: This shows that Amazon's current earnings quality must be read on two levels.
At the operating level: the cash profit is real, and operating cash flow is very strong.
At the distributable-to-shareholders level: it is not strong right now, because the enormous AI capex compresses distributable cash flow to a very low level. So it is not an "accounting-profit bubble," but neither is it a mature enterprise that "can harvest large amounts of cash right now."
Next, the balance sheet and working capital.
| Year | Cash + marketable securities | Long-term debt | Net cash | Inventory | Receivables and other current assets | Accounts payable | Shareholders' equity | Year-end shares outstanding |
|---|---|---|---|---|---|---|---|---|
| 2020 | 84.4 | 31.8 | 52.6 | 23.8 | 24.5 | 72.5 | 93.4 | 10.1 |
| 2021 | 96.0 | 48.7 | 47.3 | 32.6 | 32.9 | 78.7 | 138.2 | 10.2 |
| 2022 | 70.0 | 67.2 | 2.9 | 34.4 | 42.4 | 79.6 | 146.0 | 10.2 |
| 2023 | 86.8 | 58.3 | 28.5 | 33.3 | 52.3 | data to be added | 201.9 | 10.4 |
| 2024 | 101.2 | 52.6 | 48.6 | 34.2 | 55.5 | 94.4 | 286.0 | 10.6 |
| 2025 | 123.0 | 65.6 | 57.4 | 38.3 | 67.7 | 121.9 | 411.1 | 10.7 |
At March 31, 2026, the company's cash and marketable securities had risen further to about 143.1 billion dollars and long-term debt to about 119.1 billion dollars, leaving net cash still positive but clearly below the more comfortable state of the past. Accounts payable for 2023 was not separately pulled in this round, so it is marked "data to be added."
Further Judgment on Financial Quality
ROE / ROA / ROIC On a simplified average basis, Amazon's 2024-2025 ROE was roughly in the 22%-24% range and ROA around 10%; if operating invested capital is computed simply as "shareholders' equity + long-term debt - cash and marketable securities," 2024-2025 operating ROIC was roughly in the low 20% range. This is not an actuarial measure, but it is enough to show that, with services now a larger share of revenue, Amazon's return on capital is clearly better than that of most retailers. That said, the 2025-2026 AI build-out will make this metric more volatile over the next few years, so it cannot be extrapolated statically. The ROIC here is a simplified inferred value and would require fuller segment-asset and tax adjustments to model precisely.
Net debt/EBITDA and interest coverage At year-end 2025 the company remained in a net-cash position; into Q1 2026, despite a marked rise in long-term debt, cash and marketable securities still exceeded long-term debt. On interest coverage, looking only at Q1 2026, operating profit of 23.85 billion dollars against interest expense of 0.8 billion dollars gives a coverage ratio close to 30x, indicating that debt-servicing capacity is nowhere near worrying levels for now.
Changes in share count, dividends, and buybacks Year-end shares outstanding grew from about 10.066 billion in 2020 to about 10.731 billion in 2025, so dilution is real; the company pays no dividend and did no buybacks in 2023-2025. For long-term shareholders, this means Amazon is not a company that lifts per-share value through buybacks, but one that drives per-share intrinsic value through reinvestment. If future returns on capital fall while the share count keeps expanding, the investment appeal will drop noticeably.
Whether accounting profit matches cash flow, and whether there are accounting red flags I see no direct sign of obvious financial fraud or aggressive accounting: the company disclosed that 2025 internal controls were effective, and the auditor issued an unqualified opinion. What warrants attention is not "fraud" but the impact of estimates on profit: estimates such as the useful lives of servers and networking equipment affect AWS depreciation and profit; in 2025 the company also disclosed that changes in such estimates increased that year's depreciation and amortization by 1.4 billion dollars and reduced net income by 1.0 billion dollars. So Amazon's accounting is not untrustworthy; rather, its volatility and comparability are amplified by investment gains, fair-value changes, depreciation lives, and one-off settlement costs.
Financial Quality Conclusion If the question is "is this a real, resilient cash-flow machine," the answer is yes; if the question is "is the cash it can distribute to shareholders today already very abundant," the answer is not for now. At this stage, Amazon looks more like a "high cash generation + high cash re-consumption" enterprise.
Owner Earnings and Intrinsic Value
Owner Earnings Analysis
Fact: In 2025 Amazon had net income of 77.67 billion dollars, operating cash flow of 139.51 billion dollars, and net capex of 128.32 billion dollars; in Q1 2026 TTM operating cash flow was 148.53 billion dollars while TTM free cash flow was only 1.2 billion dollars, and management stated explicitly that the main cause was a large rise in capex due to AI investment.
The Crux of the Problem Buffett-style owner earnings should not be set simply equal to accounting net income, nor simply equal to current free cash flow. For Amazon, current free cash flow is heavily depressed by the AI build-out, while net income is distorted by investment gains, one-off costs, and accounting estimates. The real question is: how much "maintenance capex" does it actually take to hold the current competitive position. The company does not disclose this figure directly.
A Conservative Estimation Method I use a method that leans conservative but stays transparent:
Start from 2025 operating cash flow of 139.5 billion dollars.
Deduct the roughly 19.2 billion dollars of dilution-related cost from equity compensation in 2025, since SBC is a real cost to shareholders.
Then assume maintenance capex of about 55 billion to 65 billion dollars. This is an assumption, not a disclosed value; the logic is to treat a substantial part of the outsized 2025 AI build-out as growth rather than fully maintenance, while not using an extremely low maintenance assumption to flatter the economics. On this basis, 2025 conservative owner earnings land roughly in the range of 55 billion to 65 billion dollars. I lean toward setting the midpoint at about 55 billion dollars. This is below GAAP net income and well above that year's free cash flow, consistent with Amazon's current reality of "accounting profit < operating cash profit > distributable free cash flow."
Inference: Against the current market cap of about 2.88 trillion dollars, the market values Amazon at roughly 41-52x the owner earnings in the range above, with a midpoint of about 52x. That is not a cheap multiple.
My Owner Earnings Conclusion
Net income: very strong in 2025, but it includes investment and one-off factors and cannot be applied mechanically.
Operating cash flow: real and robust.
Maintenance capex: unknown, and must be assumed within a range.
Real distributable cash flow: I conservatively put it at around 55 billion dollars, with a wide range of 50-70 billion dollars.
Relationship between free cash flow and net income: over the long run the two may converge or may diverge clearly in heavy-reinvestment years; 2021-2022 and 2025 are typical divergence years.
Intrinsic Value Estimate
Owner Earnings Discount Method
This is the method I weight most heavily, but I have to admit it is highly sensitive to maintenance capex and to the return on AI investment.
| Scenario | Starting owner earnings | Growth over the next ten years | Discount rate | Terminal growth | Intrinsic value per share |
|---|---|---|---|---|---|
| Conservative | 50 billion dollars | 8% | 10% | 3% | about 99 dollars |
| Neutral | 60 billion dollars | 10% | 9% | 3.5% | about 174 dollars |
| Optimistic | 70 billion dollars | 12% | 8.5% | 4% | about 285 dollars |
The "starting owner earnings" and growth rates here are assumptions, not realized figures; the share count is estimated at about 10.751 billion shares based on the 2026 proxy.
My Interpretation
The conservative scenario assumes: AWS growth slows, AI returns are ordinary, and advertising and seller-services growth decelerates.
The neutral scenario assumes: AWS and advertising keep driving margin improvement, but AI returns are merely "good rather than spectacular."
The optimistic scenario assumes: Amazon keeps beating expectations in AI cloud and advertising, and the current enormous capex is ultimately absorbed by high returns.
Opinion: At the current price of 264.86 dollars, the stock sits roughly near my optimistic scenario. For a value investor, this is not an ideal position.
Relative Valuation Method
Fact: As of the May 19, 2026 snapshot, the trailing P/Es of AMZN, MSFT, WMT, and COST were roughly 31.7x, 25.2x, 46.5x, and 56.0x respectively.
Inference:
On P/E alone, Amazon is not more expensive than Walmart or Costco, and is even cheaper; but this does not make it cheap, because Amazon's bigger problem right now is not P/E but the fact that free cash flow is suppressed by AI spending.
Against year-end 2025 shareholders' equity of 411.07 billion dollars, the current market cap implies about 7x P/B, which is not low.
Against 2025 free cash flow of about 11.2 billion dollars, the current market cap implies about 257x P/FCF, which is extremely expensive; this shows 2025 was a typical "heavy-capex year" and that P/FCF cannot be used mechanically for valuation, but it also reminds you that the market is already paying for future cash flow, not for present cash flow.
Comparable measures of EV/EBITDA and precise ROIC were not fully pulled for every comparable company this round, so I will not draw pseudo-precise conclusions here; I can only offer a directional view: at the current price Amazon is by no means a "cheap stock" and looks more like a "high-quality premium stock."
Asset or Liquidation Value Method
Amazon is not suited to liquidation value as a primary valuation method. Fact: At year-end 2025 the company's cash and marketable securities totaled about 123.0 billion dollars and long-term debt about 65.6 billion dollars; by March 31, 2026, cash and marketable securities were about 143.1 billion dollars and long-term debt about 119.1 billion dollars. The company's net property and equipment was 357.0 billion dollars, but a large part of that is data centers and fulfillment and technology infrastructure, where the liquidation discount could be very large. Book value materially understates its platform and data assets, but liquidation value also cannot be viewed generously.
My Intrinsic Value Range
Conservative intrinsic value range: 95-125 dollars per share
Fair intrinsic value range: 150-210 dollars per share
Optimistic intrinsic value range: 250-310 dollars per share
Based on the current price of 264.86 dollars:
Relative to the conservative range, there is a premium of roughly 110%-180%.
Relative to the fair range, there is a premium of roughly 25%-75%.
Relative to the optimistic range, it sits near the range.
The Price Band I Give
Required margin of safety: at least 25%-30% below neutral intrinsic value.
Ideal buy price range: 120-155 dollars
Acceptable holding price range: 155-220 dollars
Clearly overvalued price range: above 250 dollars These are not market quotes but opinions I derive from the assumption framework above.
Expected Annualized Return
If you buy at about 264.86 dollars today and hold for 10 years, my rough view of the price return is:
Conservative scenario: about -3% to 0% per year
Neutral scenario: about 3% to 6% per year
Optimistic scenario: about 8% to 11% per year This is not a short-term forecast but a rough projection based on owner-earnings growth and the terminal valuation. Put differently, to earn a decent return buying Amazon today, you need to rely on "fairly good operating delivery."
Margin of Safety and the Bear Case
Is the margin of safety sufficient
My answer is unequivocal: insufficient.
Because the current price already embeds several conditions that "all need to hold at once":
AWS can maintain high growth and high margins in the AI era.
The current AI infrastructure investment ultimately proves to be high-return rather than a low-return capital sink.
Advertising, seller services, and subscriptions keep lifting the overall margin.
Regulation does not reshape the monetization logic of Marketplace and advertising.
If even one or two of these are clearly cut down, the current valuation becomes fragile.
The Most Fragile Assumption in the Valuation
The most fragile assumption is not "revenue keeps growing" but the quality of that growth: The market does not doubt that Amazon will keep getting bigger; what the market is really betting on is whether the incremental capital it pours into AI can ultimately sustain the high-return characteristics it historically achieved with AWS. If it cannot, Amazon could degrade from a "high-return compounding machine" into a "high-growth but increasingly capital-intensive platform company."
What happens if growth falls short, margins decline, and the multiple compresses
If growth falls short but the AWS and advertising moats stay stable, you might still earn a mediocre return, just not an outstanding one.
If margins decline, especially if AWS margins fall back clearly while capex stays high, the investment thesis takes clear damage.
If the multiple retreats from a 31.7x P/E or roughly 50x owner earnings to a more ordinary range, then even if revenue keeps growing, you could see years of "the business advances, the shareholder makes nothing."
The Most Important Risk List
Competitive risk: AWS share and technical mindshare face catch-up from Azure and Google Cloud.
Technology substitution risk: AI cloud architecture, the chip stack, in-house models, and the inference-cost curve change so fast that they could rewrite the cloud competitive landscape.
Regulatory risk: antitrust, platform rules, the Featured Offer, seller-data use, the Prime structure, and more could all be constrained.
Capex risk: if the 2025-2026 AI investment returns less than expected, it will suppress free cash flow and the valuation.
Management risk: Bezos remains an important spiritual and controlling figure, and Jassy's capability in AI capital allocation is the most critical variable over the next few years.
Overvaluation risk: the market has paid a fair amount of future-quality premium in advance.
Cyclical risk: weak consumption affects retail and advertising, and changes in corporate IT budgets affect AWS.
Supply-chain and energy risk: AI infrastructure depends on chips, electricity, and data-center build capacity.
FX and tax risk: the international business is broad, and tax disputes and currency swings both affect profit.
Accounting-comprehension risk: investment fair value, depreciation lives, and one-off settlement costs amplify profit volatility.
The Strongest Bear Case
The strongest bear case is actually very plain:
Amazon may still be a great company, but if you buy it today, what you are buying is not "cheap greatness" but "expensive excellence."
If AWS's competitive position weakens somewhat, the marginal returns on AI infrastructure fall short of expectations, and free cash flow takes longer than expected to recover, then today's valuation leaves long-term shareholders little room for error. What the bears truly see is not that Amazon will decline, but that even an excellent company can deliver mediocre returns.
Which facts would overturn my judgment
If the following facts appear in the future, I will concede I have been too conservative:
AI investment clearly converts into higher AWS growth and stronger operating cash flow within 2-3 years.
AWS share stabilizes or even recovers while margins stay high.
Advertising and seller services keep growing fast, lifting the overall operating margin to a higher level.
Free cash flow recovers quickly even amid high capex, while share dilution does not worsen.
Conversely, if the following facts appear, I will consider the bull thesis damaged:
AWS share keeps falling and margins drop below the 30% area.
Capex stays extremely high, but operating cash flow and the backlog fail to rise in step after 2-3 years.
Regulation materially weakens Marketplace/Prime/advertising monetization.
Equity-compensation dilution accelerates while per-share intrinsic-value growth slows.
The Largest Permanent Capital-Loss Scenario
It is not company bankruptcy but buying high and then having the return eroded over time by a high valuation and low cash returns. If the market ultimately re-rates Amazon as a "high-quality but more capital-heavy" platform company, rather than the increasingly asset-light, increasingly high-return compounder of the past, long-term shareholders could endure a price drawdown of 35%-55% and struggle for years to earn back the opportunity cost. That is the true "permanent capital loss" to guard against here. This is an inference, not an established fact.
Comparison, Checklist, and Final Judgment
Comparison with Other Opportunities
Versus the strongest competitor Looking only at the balance of "business certainty + current valuation," Microsoft may not be cheaper than Amazon right now, but its cloud-and-software path to cash is more direct; Amazon's retail-advertising-AWS combination is more unique, but its free cash flow is currently more suppressed by capex. Through the lens of "both are super-platforms," Amazon's advantage lies in the optionality of its business mix, and its disadvantage lies in greater uncertainty around short-to-medium-term capex.
Versus a broad-based index SPY, as an S&P 500 ETF, has the merits of diversification, simplicity, and less dependence on a single management team and a single valuation error. As a single company, if bought at the current price, I do not think Amazon clearly beats buying the index on win-rate; only at a lower price, or if you have stronger conviction in its AI return on capital, does it look more like an opportunity that clearly beats the index.
Versus the risk-free rate or high-grade bonds The U.S. Treasury continuously publishes a daily yield curve, which shows the risk-free rate itself is not low. For an investor with a balanced risk appetite, the median 10-year expected annualized return implied by Amazon's current valuation is not especially generous, so the risk compensation is not abundant. I do not write a specific same-day 10-year figure here, because this round did not separately pull the precise level for the day; if you want a hard comparison against Treasuries, I suggest adding a same-day yield.
If you could hold only five assets, does it qualify for the portfolio
It qualifies on business quality; on price, I am abstaining for now. In other words: at the company level it is a top-five candidate, but at the price level it is not.
Investment Checklist
| Check Item | Conclusion | Note |
|---|---|---|
| Can I understand this business | Pass | Compounding, but the core flywheel is clear. |
| Does it have stable long-term demand | Pass | Long-term demand in e-commerce, cloud, and advertising remains. |
| Does it have a durable moat | Pass | Scale, network, data, fulfillment, AWS. |
| Does it have pricing power | Partial pass | AWS/advertising/subscriptions are strong, retail is weak. |
| Can it generate stable free cash flow | Fail | At this stage free cash flow is strongly suppressed by AI capex. |
| Is its return on capital excellent | Pass | Simplified ROIC has been very strong the past two years, but the future is yet to be proven. |
| Is management trustworthy | Pass | Long-term oriented, with fairly candid disclosure. |
| Is capital allocation rational | Uncertain | Historically excellent; the AI return is not yet settled. |
| Is the balance sheet sound | Pass | Strong liquidity, manageable debt. |
| Is the valuation below intrinsic value | Fail | My neutral valuation is below the current price. |
| Is the margin of safety sufficient | Fail | Not obvious. |
| Does long-term holding put me at ease | Conditional pass | At ease at a reasonable price; the current price is not comfortable enough. |
| Which key facts would make me sell | Needs ongoing tracking | AWS share/margins, AI capex returns, regulation. |
| Am I only wanting to buy because the price rose or because of market sentiment | Fail | If I wanted to buy now, I think I would easily be swayed by the "good company narrative." |
Final Investment Conclusion
【Final Rating】 Watch
【One-Sentence Investment Thesis】 Amazon remains one of the finest compounding platform companies in the world, but the current price is closer to "paying in advance for future excellence" than to "a worthwhile buy even under conservative assumptions."
【Core Bull Case】
Retail traffic, the seller ecosystem, Prime, advertising, and AWS form a multi-layered flywheel, with a deep and rare moat.
2025 operating profit and operating cash flow both hit highs, showing the underlying business quality is extremely strong.
AWS and advertising are high-margin engines, with long-term room to grow.
The balance sheet remains sound and can withstand intense AI investment and cyclical swings.
Management's long-term orientation is distinct, with an excellent historical track record in capital allocation.
【Core Bear Case】
The current valuation demands high returns on future AI investment, and the margin of safety is insufficient.
Free cash flow is clearly compressed by heavy capex in 2025-2026.
AWS share faces catch-up from Azure and Google Cloud.
Regulatory and antitrust risk persists and could affect core platform rules.
Equity compensation causes ongoing dilution, and the company has not used buybacks to offset it in recent years.
【Key Assumptions】
AWS can hold its industry-leading position for the next several years.
The return on AI infrastructure investment is not materially worse than the historical return on AWS investment.
Advertising, subscriptions, and seller services keep driving margin improvement.
Regulation does not materially break the Marketplace and Prime flywheel.
【Fair Buy Price】 The more comfortable range I give is 120-155 dollars; if the standard is loosened, 155-180 dollars is where it begins to approach a range one could seriously act on. The basis is that this range is closer to my neutral intrinsic value and leaves a 25%-30% margin of safety.
【Target Holding Period】 At least 10 years or more. Amazon is not a stock to bet on through 1-2 year valuation mean reversion; it is better judged on "long-term platform value + return on reinvestment."
【Expected Annualized Return】
Conservative: -3% to 0% per year
Neutral: 3% to 6% per year
Optimistic: 8% to 11% per year This is a long-term projection from the current price, excluding dividends, since the company pays none.
【Maximum Loss Risk】 If AWS competitiveness weakens, AI capex returns fall short, and the multiple retreats at the same time, a medium-to-long-term price decline of 35%-55% is not unimaginable. The worst case is not the company disappearing, but your long-term return being severely diluted after buying at too high a price.
【Tracking Metrics】 Going forward I will keep watching the following metrics:
AWS revenue growth
AWS operating margin
Advertising revenue growth
Third-party seller services revenue growth
Operating cash flow and TTM free cash flow
Net capex and the capex/revenue ratio
Changes in share count and SBC expense
Prime/deferred revenue and AWS unsatisfied obligations
North America and International retail segment margins
Progress on major regulatory cases.
【Signals That Trigger Reassessment】
AWS growth lags the industry clearly for several consecutive quarters while margins keep falling.
AI capex keeps surging, but operating cash flow and the backlog show no corresponding improvement.
Regulation forces Amazon to change platform ranking, seller-data use, or key Prime terms.
Dilution picks up while per-share operating cash flow fails to rise in step.
One-off investment gains keep heavily distorting net income, making the real operating quality hard to discern.
【Reasons Not to Buy】 This is the most important part of the report: I am not buying, not because Amazon is not a good company; on the contrary, it is because it is so easy to let the price slip your mind. To buy at the current price, you make very high demands of the next five to ten years: high growth, high return on capital, high margins, and no major regulatory trouble. Such a combination does not meet my requirement for a "margin of safety."
【Final Recommendation】 If you are a long-term value investor, putting Amazon on a high-priority watch list and tracking its AWS competitiveness, AI return on capital, and cash-flow recovery is more rational than rushing in at the current price. My recommendation is not to dismiss this company but to hold to price discipline: first concede that it is excellent, then concede that it is not cheap right now.
Boundaries of the Material and Open Questions This report has already covered the most important official sources and the current price, but two boundaries remain:
I did not fully pull same-basis P/B, EV/EBITDA, and ROIC for every comparable company this round, so the relative-valuation part relies mainly on Amazon's own figures and the P/E comparison with its most important peers.
Maintenance capex is not a company-disclosed item, so the owner-earnings estimate must use a range assumption. These two points do not change my core conclusion: Amazon is an excellent company, but right now it looks more like a watch than an obvious buy.
This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.
Full report
Sign in to read the full report
Sign up free to unlock the full text, the Baillie growth scorecard, and full-text search.
Log in / Sign up free