Segment Disclosures: Finding Where the Moat Actually Lives

Fifty-seven percent. That was the share of Amazon's 2025 operating income earned by Amazon Web Services, according to Amazon's 10-K for fiscal 2025, from a segment that brought in just 18% of the company's sales. The other 82% of revenue, the North America and International retail businesses, produced the remaining 43% of the profit between them. One company, one ticker, one moat rating. Two very different economic engines underneath.

So the structural question for any business that reports more than one segment isn't "does this company have a moat?" It's narrower and more useful: which part of the company is the moat protecting, and does the rest of the business add to that advantage or dilute it? A consolidated income statement can't answer that. It blends a high-return franchise and a low-return operation into one margin that describes neither of them. The segment note can, at least roughly.

I want to work through two real cases from the most recent annual filings, Amazon and Disney, because they show the same pattern in industries that have almost nothing else in common. Then I'll lay out the reading order I use on any segment note, and where I think that method gets less reliable than it looks.

One Rating, Several Businesses

Most moat frameworks, ours included, grade a company as a whole: wide, narrow, or none. That's the right unit for an owner of the stock, since you can't buy the profitable segment alone. But it's the wrong unit for analysis. Competitive advantage is a property of a product in a market, facing specific rivals and specific customers. A conglomerate can be wide-moat in one market and openly competitive in the next, and the blended number will hide both facts.

US public companies disclose segment revenue and a measure of segment profit in a note to the annual financial statements, using the units management itself uses to allocate resources. That's the useful part. The segments aren't drawn by an outside analyst; they reflect how the people running the business think about it. Three questions get you most of the way:

• Where does the operating income sit, relative to where the revenue sits? A segment that earns a much larger share of profit than of sales is the first candidate for the advantage.
• Is that segment's margin stable across several years, including a bad one? Durable economics show up as a margin that holds, not one that spikes.
• Does the lower-margin segment strengthen the high-margin one, or merely sit next to it? Shared customers, shared infrastructure, or shared intellectual property can make a weak segment strategically necessary. Without those links, it's just dilution.

None of this requires accounting forensics. It requires two lines per segment, revenue and operating income, for three or more years. If you've worked through margin analysis before, our guide to gross margin covers the same instinct at the company level; this is the segment-level version of it.

Amazon: Where the Profit Pool Actually Sits

Amazon reports three segments: North America, International, and AWS. The first two are mostly retail (owned inventory, third-party seller services, subscriptions, advertising), split by geography. AWS sells computing, storage, and related services to businesses and governments around the world.

What the three segments earned

Here are the segment figures from Amazon's 10-K filings for fiscal 2023 through fiscal 2025. Margins are segment operating income divided by segment net sales, which I've calculated myself:

• AWS: net sales of $90.8 billion in 2023, $107.6 billion in 2024 and $128.7 billion in 2025. Operating income of $24.6 billion, $39.8 billion and $45.6 billion. Margin: 27.1%, 37.0%, 35.4%.
• North America: net sales of $352.8 billion, $387.5 billion and $426.3 billion. Operating income of $14.9 billion, $25.0 billion and $29.6 billion. Margin: 4.2%, 6.4%, 6.9%.
• International: net sales of $131.2 billion, $142.9 billion and $161.9 billion. Operating income of −$2.7 billion, $3.8 billion and $4.8 billion. Margin: −2.0%, 2.7%, 2.9%.

The three segments add up exactly to Amazon's consolidated figures: $716.9 billion of net sales and $80.0 billion of operating income in 2025, a blended margin of 11.2%. That 11.2% describes no part of the business. AWS earns roughly three times it. The two retail segments together earned 5.8% in 2025, about half of it.

Segment share of profit ÷ segment share of revenue — AWS, fiscal 2025: (45.6 ÷ 80.0) ÷ (128.7 ÷ 716.9) = 57.0% ÷ 18.0% ≈ 3.2×

A ratio well above 1 doesn't prove a moat, but it tells you where to start looking. At 3.2×, it's hard to miss which segment has pricing power the others lack.

Why AWS holds its margin

A high margin is a result. The moat question is whether something structural defends it. What would it take to displace AWS from a customer that already runs its workloads there? Quite a lot. Applications are built around a provider's specific services, data is expensive and slow to move, and the engineers who run it have trained on one set of tools. That's the switching-cost mechanism described in our post on switching costs as a moat source, and it applies to infrastructure that a customer's own products depend on. Add the scale of the data-center footprint, which spreads fixed costs over a very large revenue base, and you have two moat sources reinforcing each other.

The margin history matters as much as the level. AWS earned 27.1% in 2023, a year when many cloud customers were actively cutting their spending, and stayed above 35% in both of the following years. A segment that keeps a margin in the high twenties during its weakest stretch is showing you that its pricing isn't hanging on a single good year.

What the retail segments are actually telling you

The retail segments are harder to read, and I'll admit I'm less sure about this half of the analysis than I'd like to be. On the segment numbers alone they look like scale businesses with thin returns: a 6.9% margin in North America and 2.9% internationally in 2025, after International lost money as recently as 2023. That doesn't look like a wide moat on its own.

But segment boundaries can hide a second profit pool. Amazon's advertising business sits inside the retail segments; the company discloses advertising revenue but not a separate profit figure for it. So part of the retail margin improvement from 2023 to 2025 could come from a high-margin business that depends on retail traffic rather than from the retail operation getting better at its core job. The segment note can't settle which. It can tell you that the question exists, and that's worth something.

A fair reading, I think: AWS carries the clearest evidence of a wide moat; the retail segments show a narrow advantage at best on these figures, with the honest caveat that their reported margin blends in businesses that look nothing like retail.

Disney: The Same Pattern in a Different Industry

Disney has three segments: Entertainment (studios, television networks, streaming), Sports (mainly ESPN), and Experiences (theme parks and resorts, cruise ships, consumer products). Its fiscal year ends around the end of September, so the figures below are for the years ended September 2023, September 2024 and September 2025, from Disney's 10-K filings for those years.

Experiences as the load-bearing segment

• Experiences: revenue of $32.5 billion, $34.2 billion and $36.2 billion. Segment operating income of $9.0 billion, $9.3 billion and $10.0 billion. Margin: 27.5%, 27.2%, 27.6%.
• Entertainment: revenue of $40.6 billion, $41.2 billion and $42.5 billion. Segment operating income of $1.4 billion, $3.9 billion and $4.7 billion. Margin: 3.6%, 9.5%, 11.0%.
• Sports: revenue of $17.1 billion, $17.6 billion and $17.7 billion. Segment operating income of $2.5 billion, $2.4 billion and $2.9 billion. Margin: 14.4%, 13.7%, 16.3%.

In the year ended September 2025, Experiences produced 37.5% of segment revenue (before intersegment eliminations) and 56.9% of total segment operating income of $17.6 billion. Same signature as AWS: a profit share well above the revenue share. And look at the steadiness. The Experiences margin moved within half a percentage point across three years. Few segments anywhere are that consistent.

The structural reason isn't hard to name. A destination theme park needs land, many years of construction, and, crucially, characters and stories that families already care about. A rival can raise the capital. It can't raise decades of attachment to the intellectual property. That's an intangible-asset moat (see our post on intangible assets as moats) protecting a physical business, which is an unusual and very durable combination.

Entertainment's recovery isn't the same as a moat

Entertainment's margin tripled, from 3.6% to 11.0%, over the same period. A screen would flag that as the exciting segment. But a margin rising from a depressed level tells you about cost discipline and a business cycle, not about whether competitors can be kept out, and the streaming market Disney competes in is crowded with well-funded rivals. Consider what the segment actually does for the moat: it creates the characters that fill the parks. Entertainment may be a narrow-moat business on its own and still be the supply line for the wide-moat one. That's the link to keep in mind. A weak segment that feeds the strong one is a very different thing from a weak segment that just happens to share a holding company.

A Reading Order for Any Segment Note

Here's the sequence I run, in this order, on any company reporting two or more segments:

1. Pull revenue and operating income for each segment over at least three fiscal years. Check that they reconcile to consolidated totals, allowing for eliminations and corporate costs. If they don't come close, find out why before going further.
2. Compute each segment's share of profit divided by its share of revenue. Anything well above 1 is a candidate for the advantage; anything well below 1 is diluting it.
3. Check margin stability in the candidate segment, especially in the weakest year in the window. A stable margin through a soft year is better evidence than a high margin in a strong one.
4. Name the mechanism. Switching costs, network effects, intangibles, cost advantage, efficient scale: if you can't say which one defends the segment's margin, the margin may be cyclical.
5. Ask what the low-margin segments do for the high-margin one. Feeding customers, IP or infrastructure to the moat is a reason to keep them. Merely sharing a ticker is not.

This also feeds the company-level screening approach in our guide to finding wide-moat stocks. A company that passes a consolidated return screen deserves a second look at the segment level, to see whether the returns are broad-based or rest on a single segment.

Where the method breaks

Segment operating income depends on how a company allocates shared costs, such as technology, corporate overhead and stock compensation, and the filings describe those methods only in general terms. Two companies with identical economics can show different segment margins because of allocation choices. Companies also redraw segments from time to time; Disney's current three-segment structure dates only from its 2023 reorganization, which limits how far back you can compare like with like. And as the Amazon advertising example shows, the most interesting profit pool is sometimes inside a segment rather than reported as one. Treat segment margins as strong evidence about direction, not as precise measurements.

I'd also caution against reading a high-margin segment as an automatic case for owning the whole company. Our post on asset-light business moats argues that capital intensity changes what a margin is worth, and that applies segment by segment. AWS and Disney's parks are both heavy on capital. Their margins are real; whether they're enough to justify the capital behind them is a separate question.

Key Takeaways

• A company-wide moat rating can hide a business in which one segment carries the advantage and the rest operate in plainly competitive markets. The segment note in the annual report shows you which is which.
• Compare each segment's share of operating income with its share of revenue. In fiscal 2025, AWS earned 57.0% of Amazon's operating income on 18.0% of sales; Disney's Experiences earned 56.9% of segment operating income on 37.5% of segment revenue.
• Margin stability through a weak year is better evidence of durable economics than a high margin in a strong one. Disney's Experiences margin stayed between 27.2% and 27.6% over three fiscal years.
• A low-margin segment that supplies the moat (Disney's Entertainment feeding characters to the parks) is a different case from one that simply dilutes it. Ask which you're looking at before writing either off.