Report contents
Industry landscape
Where the company competes: a national retail franchise and a global cloud oligopoly
Industry structure: concentrated in cloud, fragmented almost everywhere else. Amazon reports three segments — North America, International, and Amazon Web Services. Fiscal 2024 net sales were $638.0 billion, split $387.5 billion in North America, $142.9 billion internationally, and $107.6 billion at AWS, per the company's 10-K. Those are not variants of one business. The retail segments sell goods and marketplace services to consumers country by country. AWS rents computing capacity by the hour to businesses worldwide. Advertising, disclosed inside the retail segments rather than as its own line, sells placement on pages Amazon already owns. Each arena has a different set of rivals and a different answer to the question of whether scale protects returns.
The retail value chain starts upstream with manufacturers and millions of independent sellers and ends downstream with consumers who buy a single unit at a time. Amazon buys and resells first-party inventory and also collects commissions, fulfillment fees and ad dollars from third-party sellers, who supply roughly 60% of paid units. On the same order, Amazon is merchant and landlord to the seller. Upstream suppliers are fragmented and downstream buyers are fragmented, which is unusual in retail and is the mechanical reason Amazon can earn a return on the fulfillment network that neither side can capture for itself. That structure is national, not global. Prime membership, one-day delivery density and ad inventory do not travel across borders cheaply; a warehouse in Kentucky does nothing for an order in São Paulo.
Geographic scope therefore differs by segment. US online retail is a national market with one dominant player. International retail is a patchwork of country markets where Amazon is one competitor among several: it ranks among the largest online retailers in Germany, the UK and Japan, competes with Mercado Libre for leadership in Brazil and Mexico, closed its Chinese domestic marketplace in 2019, and faces Flipkart and Reliance in India. Cloud is the opposite: AWS sells standardized compute and storage globally, with data-residency rules creating regional sub-markets rather than separate competitive arenas. Advertising is global in reach but monetizes US retail traffic most heavily, which ties its growth to the retail franchise.
| Arena | Amazon's position | Largest rivals | Concentration |
|---|---|---|---|
| US online retail | about 40% of US e-commerce (estimate) | Walmart about 7%; Apple, eBay, Target smaller | Amazon plus four others take most online sales, but total US retail, Amazon included, is fragmented |
| Global cloud infrastructure | AWS about 30% (estimate) | Microsoft Azure about 20%; Google Cloud about 12% | three firms take roughly two-thirds of spend |
| Global digital advertising | about 9% (estimate) | Google and Meta together near half | top three take roughly 60% |
Retail and advertising shares are estimates published by eMarketer; the cloud figures are Synergy Research estimates. Amazon's own filings confirm the segment structure but do not disclose share.
The practical test for entry barriers is whether the leaders can be counted on one hand. Cloud passes it: AWS, Microsoft and Google have held the top three positions for a decade, and the capital required to match them — servers, land, power contracts and a decade of software — is now measured in the tens of billions annually. Advertising passes it in structure but not in Amazon's favor: two firms take about half of global digital ad spending, and Amazon's third-place position rests on retail traffic that neither can easily replicate, though it also means Amazon is the challenger rather than the incumbent. US online retail passes it only if the arena is defined as online sales. Widen the definition to all US retail and Amazon's share falls to roughly 8%, and Walmart remains the larger merchant by revenue. The barrier in retail is specific to the online channel and to fulfillment density, not to retail generally.
That distinction shapes the whole assessment. Amazon's advantage is strongest where a single national network serves a single national customer base, which is why North America is the segment with the longest record of high returns and why International has been slower to earn its cost of capital. It is weakest in country markets where local incumbents already own delivery density. The cloud franchise is the cleanest oligopoly of the three, and the one whose boundaries are global. The open question is whether advertising, which now sits on top of the retail network, behaves like a third concentrated business or like a tax on sellers that competitive marketplaces can undercut. So far the evidence shows share and revenue growth without telling us how durable the pricing is.
The retail segments and AWS share a balance sheet, a distribution network and a customer base, but they do not share a market. Judging Amazon as one competitive position obscures where the returns actually come from: a nationally scaled logistics and marketplace system that few can replicate inside the US, and a globally scaled cloud platform that two deep-pocketed rivals are actively contesting.
Competitive advantage
Does the company have a moat? Yes, concentrated in cloud and US retail
Moat rating: Strong, in two arenas rather than across the company. Amazon holds the leading position in two markets where entry is genuinely hard: US online retail, at roughly 40% of e-commerce on eMarketer estimates, and global cloud infrastructure, at roughly 30% on Synergy Research estimates. Both have generated returns above the cost of capital through the past five fiscal years. The protection is not uniform. In international retail Amazon is one competitor among several, and in digital advertising it trails Google and Meta with about 9% of global spend. Two tests decide whether an advantage is structural: whether the leaders stay few, and whether excess profits persist instead of being competed away. Amazon passes both in cloud infrastructure and US online retail. It does not pass either in country-by-country retail outside North America.
Market position: first in two arenas, mid-pack in many others
Market position: Strong. In the arenas that drive profit, Amazon ranks first and by a wide margin. US online retail puts it ahead of Walmart, Apple and eBay, and the gap is large enough that no rival has closed it in five years. Global cloud infrastructure makes AWS the largest provider ahead of Microsoft Azure and Google Cloud, a top three that has not changed membership in a decade. Advertising has grown into a third business inside the retail segments without becoming a third concentrated position, because the two leaders together take close to half of global digital ad spending. International retail is where the ranking breaks: Amazon closed its Chinese domestic marketplace in 2019, competes with Mercado Libre for leadership in Latin America and with Flipkart in India. The share figures for retail and advertising are eMarketer estimates and the cloud figures are Synergy Research estimates; Amazon itself does not disclose share in any market.
Share stability: rankings held, verified series are thin
Stability: Fairly Strong. No rival has displaced Amazon from the top of US online retail or cloud infrastructure in the past five fiscal years, and revenue rose every year in every segment. That is suggestive, not conclusive, because the arenas themselves grew.
| Arena and source | FY2021 | FY2022 | FY2023 | FY2024 | FY2025 |
|---|---|---|---|---|---|
| US online retail, Amazon (eMarketer estimate) | ~40% | ||||
| Global cloud infrastructure, AWS (Synergy Research estimate) | ~30% | ||||
| Global cloud infrastructure, Microsoft Azure (Synergy estimate) | ~20% | ||||
| Global cloud infrastructure, Google Cloud (Synergy estimate) | ~12% | ||||
| Global digital advertising, Amazon (eMarketer estimate) | ~9% |
Blank cells mark data that is not published or not verifiable. Amazon's filings carry no share series, and the third-party estimates available cover the current year. The standard test here is whether average share moves more than 5 percentage points over five years, which would point to absent barriers, or less than 2 points, which would point to very strong ones. That test cannot be run on the public record, so the rating rests on rankings and revenue instead, with the caveat attached.
Profitability: above the cost of capital, unevenly and only recently
Profitability: Fairly Strong. Amazon does not disclose return on invested capital, so return on equity is the substitute.
| Fiscal year | Revenue (USD bn) | Operating margin | Net margin | ROE | Free cash flow (USD bn) |
|---|---|---|---|---|---|
| 2021 | 469.8 | 5.3% | 7.1% | 28.8% | -14.7 |
| 2022 | 514.0 | 2.4% | -0.5% | -1.9% | -16.9 |
| 2023 | 574.8 | 6.4% | 5.3% | 17.5% | 32.2 |
| 2024 | 638.0 | 10.8% | 9.3% | 24.3% | 32.9 |
| 2025 | 716.9 | 11.2% | 10.8% | 18.9% | 7.7 |
The five-year average ROE of about 17.5% sits in the 15%–25% band that marks a competitive advantage, but the average hides a $2.7 billion loss in 2022 and a sharp recovery since. Gross margin climbed from 42.0% in 2021 to 50.3% in 2025, which is a mix shift toward AWS, advertising and third-party seller services, not a uniform improvement in retail economics. An estimate built from reported figures puts fiscal 2025 after-tax operating profit near $64 billion against invested capital of about $325 billion, close to 20%, before any adjustment for operating leases. The second quarter of fiscal 2026 complicates the read: net income of $62.65 billion exceeded operating income of $27.46 billion by more than $35 billion, a gap that implies a large non-operating gain, so the operating line is the cleaner measure of the business.
Overall: Strong, with the moat doing its work in two places
Overall rating: Strong. Market position rates Strong on the two leading franchises, stability rates Fairly Strong, and profitability rates Fairly Strong on returns that clear the cost of capital but swing widely. Because profitability reaches the higher band, the overall rating can sit above it. The rating would weaken if cloud share slipped toward the mid-teens, if operating profit stopped growing faster than the capital invested in it, or if a regulatory remedy changed how Amazon treats the sellers who depend on its marketplace. Amazon does not disclose Prime membership or churn, and the absence of that figure is the single largest gap in this assessment.
Where the advantage comes from and how long it can last: density, captivity and a capex test
Amazon's protection rests on three mechanisms, and none of them is a patent, a license or a scarce mineral. They are scale in physical delivery and in cloud infrastructure, customer captivity that runs deeper in cloud than in retail, and a two-sided marketplace whose seller supply reinforces itself. Each can be attacked with money, and rivals are currently attacking all three.
Supply-side scale: Fairly Strong
The cost advantage comes from density, not from size in the abstract. A fulfillment network is mostly fixed cost: buildings, robots, line-haul, delivery stations. Amazon's 2025 annual report puts more than one million robots in its fulfillment centers handling stowing, picking, sorting and internal transport. Once those costs are sunk, the economics improve with the number of packages per route, because the marginal cost of adding one more stop to a truck already on the street is small. More volume per zip code lowers cost per package, which funds faster delivery at a price rivals with thinner routes cannot match without losing money. Faster delivery lifts conversion, which adds volume, which raises density again.
The rural build-out shows the loop running in the hardest geography. Amazon has committed over $4 billion to expand rural delivery and expects the finished network to reach more than 13,000 zip codes; monthly same-day customers in rural areas nearly doubled in 2025, the company said in that report. Competitors have been pulling back from those routes, which widens the density gap where it is most expensive to close.
Cloud scale works differently but compounds the same way. Building a database service, a compute instance or a machine-learning tool carries a fixed development cost that barely rises with the number of customers who use it. The 2025 letter describes an early EC2 that launched as a single instance type in one availability zone and now sits under hundreds of services. Each service built for the widest customer base can be resold to that base at almost no incremental cost, and the largest provider spreads those fixed costs furthest. The rating stops short of Strong because capital can replicate this: the entry ticket is now tens of billions of dollars a year, which is high but not prohibitive for Microsoft, Google or a determined sovereign buyer.
Demand-side captivity: Fairly Strong in cloud, weaker in retail
Amazon does not disclose Prime membership or churn, so the renewal behavior behind the subscription cannot be verified. What can be observed is the design. Prime bundles shipping speed with video, music and pharmacy discounts, and the annual fee is sunk at the start of the term, which pushes members toward using Amazon more often to justify the cost. Habit does the rest: repurchase is one click on a stored card against a stored address.
Cloud captivity is the more durable of the two. Moving a working application means re-platforming, retraining staff, re-certifying security and absorbing the operational risk of a cutover, and the larger the data store, the worse the economics of leaving. Enterprise buyers selecting a cloud provider face high search costs because the choice is long-lived and hard to evaluate in advance. The price test matters most: rivals have competed on price for a decade, and AWS has kept the largest share on third-party estimates.
Retail captivity is partial. Prime members still compare prices, and Temu and Shein have shown that a meaningful slice of shoppers will trade delivery speed and seller trust for a lower price on commodity goods. Bundle loyalty holds for convenience purchases, not for everything in the catalog.
Two-sided marketplace: Moderate
Buyers draw sellers, and third-party sellers, who supply roughly 60% of paid units, draw buyers. That loop also funds advertising, because sellers bid against each other for placement on pages Amazon controls, and it explains why ad revenue scales without Amazon building a separate audience. The limit is scope. The network is national, not global, and the same merchants sell on Walmart's marketplace, TikTok Shop and Temu, so seller supply is not exclusive. A rival with traffic can recruit the identical seller list.
Durability: the substitutes come from price, video and AI
The threat to retail is not another department store. It is a cheaper way to buy the same goods, a different surface where shoppers discover them, and a different system that answers their questions before they reach Amazon's search bar. Temu and Shein attack on price and have grown quickly, which is direct evidence that the price-sensitive end of the catalog is contestable. Short-video commerce moves discovery off Amazon's pages, and an AI assistant that recommends a product can capture the decision before a search happens. That last one matters more than it looks: Amazon's ad inventory is valuable because shoppers start their search inside Amazon. If the starting point shifts, the ad slot loses value. In cloud, the substitutes are open-source models customers host themselves, large buyers building their own capacity, and aggressive pricing from the other two hyperscalers.
Regulation adds a second front. The FTC filed a monopolization suit against Amazon in September 2023, and the case remained pending as of early November 2026; the EU has designated Amazon a gatekeeper under the Digital Markets Act. Neither has yet forced a change in how the marketplace charges sellers, and the outcome is not predictable. A remedy that opened the buy box or restricted self-preferencing would hit commissions and ad revenue directly.
What would change this assessment
Watch AWS. Third-party estimates below the mid-20s for two consecutive years, or cloud margins compressing while rivals cut prices, would say the strongest leg of the moat is eroding. Watch the returns on the capital being spent now. Operating cash flow was $139.5 billion in fiscal 2025 and free cash flow just $7.7 billion, and the first half of fiscal 2026 ran about $27 billion negative on free cash flow. That spending is defensible only if operating profit grows faster than invested capital over the next three to four years. So far the evidence only shows the investment, not its return.
Generated by DeepVal AI. Data and models may contain errors or be outdated. For research only; not personalized investment advice.
Continue reading the full report
- Comparison and summaryHow Amazon.com Inc compares with its closest rivals, and our moat rating
Free to sign up · 20 research points · unlocking this report uses 10
20 points cover 2 full reports. No card needed.
Research reflects its generation date. Preview the first two modules.
More on Amazon.com Inc
Latest earnings · Q3 FY2026 · Oct 29, 2026
The profit headline fell; operating profit did not
Read the breakdownAmazon.com Inc
Business, moat and key questions
Open company page