The word "moat" gets thrown around so much in value-investing content that it's lost most of its meaning. Almost every analyst pitch starts with "this company has a strong moat." Almost no analysis explains which kind of moat, how durable it is, or what specifically would erode it.
Buffett's actual moat framework, refined over 50 Berkshire annual letters, has five distinct types. Each one has a different durability profile. Each one is breakable in different ways. And most modern compounders have at least two stacked, not one in isolation.
This post is the framework we use to compute the "moat" pillar score inside invest-like's Buffett Brain - the same logic Morningstar's Pat Dorsey codified in The Five Rules for Successful Stock Investing, applied to specific 2026 tickers so you can see what each pattern looks like in practice.
What "moat" actually means in Buffett's framework
A moat is a structural feature of a business that lets it earn returns on capital ABOVE the cost of that capital, for an extended period, without being competed away.
Three parts matter:
- Returns on capital ABOVE cost of capital - if a business earns 8% and capital costs 8%, there's no moat to defend. The math is neutral.
- For an extended period - one quarter of high returns is luck. Ten years of high returns is structure.
- Without being competed away - capitalism's default state is competition eroding profit. If a business holds high returns in equilibrium, something is stopping competitors.
That "something" is the moat. Five flavors exist. We'll walk each one with the cleanest modern example.
Pillar 1: Network effects
The simplest moat to recognize: the product gets more valuable as more people use it.
A telephone with one user is useless. A telephone connected to 100 million users is invaluable. Same product, same per-unit cost - the value comes entirely from the network.
Network effects are the most durable moat because they're self-reinforcing. Bigger network attracts more users, more users make it bigger. Breaking in requires a competitor to either subsidize early users at massive loss (Uber's $20B in losses to break taxis) or find a niche the incumbent doesn't serve.
Modern example #1: Visa (V) - 4 billion cardholders connect 100M merchants. A new payment network needs both sides to defect simultaneously. Visa has fought off Apple Pay, Google Pay, BNPL, and crypto so far - each "killer" eventually integrated with Visa rather than replacing it.
Modern example #2: Microsoft Teams + Office 365 (MSFT) - your company picks Teams partly because every other company picks Teams. The network is "companies you'll have meetings with." Slack and Zoom both run real businesses, but corporate IT defaults to the network everyone else is on.
Modern example #3: NVIDIA's CUDA ecosystem (NVDA) - this is the underappreciated one. CUDA isn't a chip; it's a software developer network. 4 million developers write code that only runs on NVIDIA hardware. AMD has competitive silicon. The CUDA network is what makes the chips unreplaceable.
Breakability: network effects break when (a) the network becomes table-stakes commodity (email is global but no provider has pricing power) or (b) a regulator forces interop.
See how Buffett Brain scores Visa's network moat - one of the clearest examples in the universe.
Pillar 2: Switching costs
Customers pay so much in time, money, or risk to leave that they stay even when alternatives are technically better.
The classic example is enterprise software. Switching your company's ERP from SAP to Oracle takes 18 months, costs $20M, breaks training, and risks bricking finance for a quarter. Even if Oracle is 30% cheaper, the business case to switch fails on risk-adjusted math.
Switching costs are subtler than network effects - the customer often doesn't see them until they try to leave. But they generate some of the highest sustained margins in equity markets.
Modern example #1: Adobe (ADBE) - the creative industry trains on Adobe products. A new designer can't take a job at a studio if they don't know Photoshop. So studios won't switch even if Figma is better at any given subtask, because their hiring pool is Adobe-trained. Subscription model locks in the cash flow.
Modern example #2: Intuit's QuickBooks (INTU) - small businesses don't switch accounting software. Period. The data migration is painful, the tax integrations break, the bookkeeper has to retrain. Intuit's annual churn is in the low single digits despite alternatives existing.
Modern example #3: TSMC (TSM) - chip designers tape out for a specific foundry's process node. Moving a design from TSMC's N3 to Samsung's 3nm is a 12-month re-engineering project. Apple has zero realistic option to leave TSMC for at least three product cycles.
Breakability: switching costs break when a credible alternative emerges that ALSO solves the switching cost. Salesforce broke Siebel by offering migration tooling AND a cheaper price simultaneously.
Pillar 3: Intangible assets (brand, IP, regulation)
The slipperiest pillar because "intangible" is intentionally vague. Three sub-types worth distinguishing:
Brand (preference + premium)
Brand only counts as a moat if it generates measurable pricing power. A famous brand that can't charge more is just marketing.
The test: would the product sell at a higher price than a comparable generic version? If yes (Coca-Cola vs supermarket cola, Hermès vs Coach), it's a moat. If no (most apparel, most consumer electronics outside Apple), it's brand awareness without economic value.
Modern example: Hermès (RMS.PA) - Birkin bags trade at 5-10× the cost of comparable luxury leather. Hermès chooses not to expand production. The brand is the moat. No technology disrupts it.
Intellectual property (patents, regulatory exclusivity)
Pharmaceutical patents and FDA exclusivity are the cleanest examples. The patent generates returns above cost of capital for the patent life. After patent expiry, the moat collapses.
This is why drug-company moats are time-bounded - you can model the exact year the moat ends. Software patents are weaker because they're hard to enforce internationally.
Modern example: Eli Lilly (LLY) GLP-1 patents - Mounjaro and Zepbound are patent-protected through at least 2036 in major markets. Until then, Lilly can price freely. After expiry, generic competition collapses pricing power overnight. The moat is durable but has an end date.
Regulatory licenses
Some businesses have a moat because the government won't let competitors in. Toll roads, alcohol distributors, certain telecom spectrum, banking charters.
Modern example: Moody's (MCO) - one of three globally recognized credit rating agencies. The "globally recognized" status is regulatory - Moody's, S&P, and Fitch are designated as Nationally Recognized Statistical Rating Organizations (NRSROs) by the SEC. New entrants don't get the designation. Net effect: three companies share a 100%-margin industry.
Pillar 4: Cost advantages
The business produces or delivers a comparable product at a lower cost than every competitor. Three sub-flavors:
Scale economies
Fixed costs spread over more units = lower per-unit cost. Costco buys at volumes Walmart can't match in specific SKUs. Amazon Web Services builds data centers at scale no competitor can match.
Modern example: Costco (COST) - the membership model converts fixed-cost overhead into a separate revenue stream. Margins on goods can sit near zero because membership fees cover overhead. No standard retailer can match this structure without rebuilding the business model.
Process advantages
The company does something operationally that competitors can't replicate. Toyota's production system. Trader Joe's private label flywheel. Southwest's point-to-point network in the 1990s.
Modern example: Toyota Motor (TM) - the Toyota Production System has been written about for 40 years and competitors still can't fully replicate it. Lean manufacturing isn't a binder you implement; it's a 50-year culture.
Unique input access
The company controls a resource competitors don't have access to. Saudi Aramco's lifting cost on oil. De Beers' historical diamond control. Certain mining concentrations.
Modern example: TSMC (TSM) - second-time appearance because it stacks. TSMC also has cost advantage from cumulative experience (yields on N3 are months ahead of Samsung, which means more good chips per wafer, which means lower per-chip cost).
Pillar 5: Efficient scale
The most overlooked moat. The market is just barely big enough for the incumbent to earn returns. A new entrant would split the market and neither company would earn its cost of capital.
This is structural - it doesn't depend on the incumbent doing anything especially well. It depends on the market geometry.
Modern example #1: Railroads (UNP, CSX, NSC, CP, CNI) - North America has 7 Class I railroads. There won't be an 8th. The capital cost of building a parallel railroad next to UNP's track is prohibitive, and the resulting two-rail market would generate sub-cost-of-capital returns for both. So no new entrant. The 7 incumbents split the continent geographically and earn high returns.
Modern example #2: Waste Management (WM) - garbage collection is structurally local. Once WM has a contract with a city, the marginal capital cost for a competitor to win that contract is high (truck fleet, route density, transfer stations) and the resulting margin if they win is low. So WM keeps the contract.
Breakability: efficient scale breaks when the market grows big enough to support a second player. This happened with airlines after deregulation - the route economics that supported one regional carrier started supporting two.
The stacking question
Most great businesses have at least two stacked moats. The lollapalooza effect (see Munger's mental models) applies here directly.
A short audit of which businesses stack what:
| Company | Network effects | Switching costs | Intangibles | Cost advantage | Efficient scale |
|---|
| Visa (V) | ✓ | ✓ | ✓ regulation | ✓ scale | - |
| Microsoft (MSFT) | ✓ Teams/365 | ✓ enterprise | ✓ brand | ✓ Azure scale | - |
| Costco (COST) | - | ✓ membership | ✓ brand | ✓ purchasing | - |
| TSMC (TSM) | - | ✓ design lock-in | ✓ IP | ✓ yields | ✓ |
| Mastercard (MA) | ✓ | ✓ | ✓ regulation | ✓ | - |
| Adobe (ADBE) | ✓ training | ✓ workflow | ✓ brand | - | - |
| Moody's (MCO) | ✓ trust network | ✓ embed | ✓ regulatory | - | ✓ 3-firm market |
| Union Pacific (UNP) | - | ✓ shipper lock | - | ✓ density | ✓ |
| Eli Lilly (LLY) | - | - | ✓ patents | - | - |
| Costco (COST) | - | ✓ | ✓ | ✓ | - |
| Apple (AAPL) | ✓ ecosystem | ✓ | ✓ brand | ✓ purchasing | - |
| Hermès (RMS.PA) | - | - | ✓ brand | - |
The businesses with 3+ moats stacked are the ones Buffett actually holds at scale through Berkshire.
The single-moat businesses (Eli Lilly being the cleanest example) generate huge returns when the moat holds, but you have to time the entry and exit around the patent cliff. They're not "buy and hold forever" candidates the way the multi-moat businesses are.
How to apply this
When you read any new pitch ("AMD has a great moat because of GPU performance"), force the conversation onto the five-pillar framework:
- Which pillar - network, switching, intangible, cost, or efficient scale?
- Is the pillar structural or temporary?
- Are at least two pillars stacked?
- What specific event would break each pillar?
If the pitch can't answer those four questions cleanly, the moat is rhetoric, not analysis.
invest-like's Buffett Brain computes a "Moat" pillar score that maps to this framework directly - the verdict page on every stock breaks down which moat types apply with severity ratings. The Boardroom debate engine goes further: Munger, Buffett, and Lynch each argue about which moat patterns matter most for the specific business.
What to do next
- Pick a ticker you own and run it through the Buffett-Fit scorer - read which moat types it credits and which it discounts.
- Use /compare to put two competing businesses side-by-side and see whose moat actually holds up.
- For deeper reading: Pat Dorsey's The Little Book That Builds Wealth is the canonical retail-level treatment of the same five-pillar framework.
The single best stock-picking discipline you can build is to require two stacked moats minimum before sizing into a position. That one rule alone disqualifies 80% of the universe and leaves you with the businesses worth thinking carefully about.