What it does · what it actually collects · why the reported numbers mislead
Warren Buffett owned this company once — thirty million shares in 1996 — and sold. Two years later he told his shareholders, in writing: "my decision to sell McDonald's was a very big mistake." The stake he let go would be worth north of ten billion dollars today (approx.). So this report begins where his regret ends: with the question of what, exactly, the world's best investor gave up — because it was never really a hamburger company.
Here is the trick most casual observers miss. Customers spent about $139 billion at McDonald's restaurants last year — but McDonald's Corporation reported revenue of only $26.9 billion. The gap is the business model: 95% of the 45,000+ restaurants belong to franchisees, independent owners who put up the capital, run the kitchens, and bear the labor and beef costs — while McDonald's collects a royalty of roughly 4–5% of their sales plus rent. Rent, because in most markets McDonald's owns or controls the land and building underneath the franchisee, marks up the lease, and collects it senior to everything else. That is why a "restaurant company" earns 46% operating margins — the margins of a landlord crossed with a licensor — and why the honest description of a share of MCD is: a royalty-and-rent stream on $139 billion of other people's sales, growing mid-single digits, returned to owners almost in its entirety.
"We are not technically in the food business. We are in the real estate business. The only reason we sell fifteen-cent hamburgers is because they are the greatest producer of revenue from which our tenants can pay us our rent." — Harry J. Sonneborn, McDonald's first president, to Wall Street analysts
Sonneborn said that in the 1960s and it remains the single most useful sentence ever spoken about this company. Of the ~$13.9 billion of franchise-segment profit McDonald's earns, the majority is rent, not royalties (approx. — 10-K analyses). The real estate sits on the books at decades-old cost (~$43 billion gross); third parties peg its fair value near $120 billion (approx.). Keep that in mind when you meet the balance sheet's one shock later in this report.
| Founded | 1940 (the McDonald brothers) · franchised nationally by Ray Kroc from 1955 · IPO 1965 at $22.50 |
| Sector / Industry | Consumer Cyclical · Restaurants (economically: royalties + real estate) |
| CEO | Chris Kempczinski (since 2019) · CFO Ian Borden |
| Makes money from | Franchise royalties ~4–5% of sales + RENT (senior) · ~5% company-run stores |
| System sales vs revenue (FY2025) | ~$139 B system-wide · $26.9 B reported |
| Market capitalisation | ~$188 B · 45,000+ restaurants · ~210M loyalty members (approx.) |
How a milkshake-machine vendor built the world's largest small-business network
Three people built this, and the division of labour explains the machine: the brothers invented fast food as a manufacturing process; the salesman turned it into a franchise system; the accountant turned the franchise system into a real-estate empire.
| Year | Milestone |
|---|---|
| 1940–48 | Dick and Mac McDonald open in San Bernardino; in 1948 they relaunch with the 'Speedee Service System' — a stripped menu and an assembly-line kitchen. Fast food is born as a process, not a recipe. |
| 1954–55 | Ray Kroc, a 52-year-old Multimixer salesman, visits to see why one restaurant needs eight milkshake machines — and franchises it nationally. His twist: sell one store at a time to owner-operators and prosper only when they do. |
| 1956 | Harry Sonneborn creates Franchise Realty Corp: McDonald's buys or leases the land and building, sub-leases to the franchisee at a markup, and collects rent senior to the royalty. The financial architecture is set — and never changes. |
| 1961–65 | Kroc buys out the brothers for $2.7M (famously walking away from a handshake 0.5% royalty worth billions today). Hamburger University opens. IPO April 1965 at $22.50 — 12 stock splits follow. |
| 1968–90 | The Big Mac (a franchisee's invention, 1968), drive-thru (1975), Happy Meal (1979), Moscow's Pushkin Square (1990, a 30,000-person queue). 'Billions and billions served' — the sign runs out of digits. |
| 2002–03 | The near-death: overexpansion ends in the first quarterly loss since the IPO ($343.8M, Q4 2002); the stock touches ~$12–13. Jim Cantalupo's 'Plan to Win' inverts the strategy — better, not bigger — and works almost immediately. |
| 2015–19 | Steve Easterbrook refranchises ~4,000 company stores (to ~95% franchised): reported revenue falls by design while margins march from ~31% toward 46%. Fired in 2019 over relationships with employees; when the board later learns he lied, it sues and claws back the full $105M severance — one of the largest clawbacks in corporate history. |
| 2019–26 | Chris Kempczinski: drive-thru carries COVID; 'Accelerating the Arches'; the 2024 value-perception crisis (the viral $18 Big Mac combo) answered with the $5 Meal Deal, McValue and repriced Extra Value Meals; loyalty scaled to ~210M active members; 45,356 restaurants at end-2025, targeting 50,000 by end-2027 — the fastest unit growth since the 1990s. |
Two lessons for an owner. First, the moat forgives management error — 2003 was a self-inflicted near-death, cured not by genius but by returning to the system's own economics; that is practically the definition of a durable franchise. Second, the money machine and the menu are separable: reported revenue fell for years while per-share earnings compounded, because the whole point of the Sonneborn architecture is that McDonald's Corporation prospers on the top line of its tenants, not its own. Readers who screen on revenue growth will screen this company out forever — and misread it forever.
How it works — and how confidently you can know its future
The machine, in five steps:
One honest boundary note: the system is 5/5 knowable; the customer carries two genuine question marks — a value-stung low-income consumer, and appetite-suppressing drugs reaching tens of millions of people (Part VI). Neither threatens the architecture; both can shave the growth rate the price assumes. That is why the growth dial reads 6, not 8.
The 5× gap · rent senior to royalties · a landlord's margins
FY2025 reported revenue by segment — remembering that each dollar below sits on roughly five dollars of system-wide customer spending:
Why the margins look impossible. A restaurant company earning a 46% operating margin and a 31.6% net margin is not running restaurants — it is licensing a brand and collecting rent. The franchisees absorb the industry's actual economics (labor inflation, beef prices, the E. coli shock, California's $20 minimum wage); McDonald's collects a percentage of their top line, insulated from their cost lines. The same architecture explains the growth math: management is opening ~2,600 restaurants a year toward 50,000 by end-2027, yet guides capex of only ~$3.7–3.9B (approx.) — franchisees and licensees fund most of the building. And the digital layer now matters: ~210 million 90-day-active loyalty members (approx.) spending ~$40B a year through the app — the largest first-party customer dataset in food service, quietly converting a habit business into a data business.
Brand, corners, scale, and 2,000 motivated owners
Four springs fill this one, and their durability is the whole investment case:
1 · Brand. The most valuable restaurant brand on Earth (~$42.6B, roughly a Coca-Cola of pure sign-value (approx. — Brand Finance)), the only restaurant in the global top tier of all brands, refreshed by the largest advertising pool in food service — funded, elegantly, by the franchisees' own 4% contributions. Trend: intact, with a genuine 2024 scare (below).
2 · The corners. Sixty years of first-pick locations — drive-thru-zoned lots acquired at historical prices, carried at cost, irreplaceable at any price because the corners are taken. This is the moat competitors literally cannot buy; it converts brand strength into contractually senior rent. Trend: appreciating silently.
3 · Scale. $139B of system sales buys supplier terms, ad reach and now AI/loyalty data no rival approaches. Wendy's and Burger King operate the same nominal model at a fraction of the density; the gap compounds. Trend: widening.
4 · Owner-operators. Kroc's original insight — sell one store at a time to people whose life savings are inside it — still outperforms every salaried alternative. The double edge: those same owners are counterparties with lawyers (Part XI). Trend: stable, tense.
Where it cracked, briefly. In 2024 a single viral photo — an $18 Big Mac combo at one rest-stop — crystallised three years of fast-food inflation and did measurable brand damage: traffic fell, low-income visits dropped double digits, and Q1 2025 delivered the worst US comp since the pandemic (approx.). The response was the system working as designed: corporate financed a ~15% combo repricing, relaunched Extra Value Meals nationally, and by Q1 2026 US comps were +3.9% with management crediting value items. A moat that takes a hit and repairs itself inside eighteen months is not a broken moat; it is an tested one. Score: 9 — narrower only than Coca-Cola's, and made of land.
The squeezed customer today · the GLP-1 decade tomorrow
Every durable franchise on our shelf carries one question the price turns on. Coca-Cola's was sugar; Visa's was disintermediation; McDonald's is the most literal of all: will people keep wanting this food? The question arrives in two forms.
Form one — the wallet. The CEO calls it "a two-tier economy": households above $100K are fine; below, customers skipped breakfast entirely or traded down to home cooking through 2025 (approx. — earnings calls). The value reset (Part V) bought back the traffic, but every point of discounting comes out of franchisee margins first — corporate had to subsidise the repricing — and a permanently poorer core customer would compress the system's growth arithmetic even with the architecture intact.
Form two — the molecule. GLP-1 appetite drugs are the first genuine demand-side technology risk in fast-food history. The evidence so far, honestly: users cut fast-food spending ~8% in their first six months; regular users' dinner visits run ~−6%; the aggregate industry impact to date is a fraction of a percent (approx. — industry studies) — and J.P. Morgan projects 30M+ American users by 2030, tripling today's base, with cheaper pills coming. Management reports no material impact yet and is pushing protein (chicken reached revenue parity with beef in 2024, ~$25B each system-wide (approx.); McCrispy is a $1B+ brand in 55 markets).
| The bear case | The bull case |
|---|---|
| 30M+ GLP-1 users by 2030 (approx., JPM) — an appetite tax on the exact demographic that eats here most | Impact to date ~−0.4% of dinner sales (approx.); the menu adapts (chicken at beef parity, protein push); 60 years of surviving every health panic |
| The low-income customer is structurally squeezed; value pricing squeezes franchisee economics in turn | The value reset already worked — Q1 2026 comps +3.9% on value items; loyalty (~210M members) deepens frequency cheaply |
| Beef costs at records; boycott geopolitics (the flag on the door cuts both ways); a food-safety tail (2024's E. coli) | 2,600 openings/yr toward 50,000 units — the growth is units and rent, not calories per customer; licensees fund the build |
Our read: the wallet question is cyclical and already answered; the molecule question is secular and honestly unanswerable — which is why it caps the growth dial at 6 and belongs in every quarterly check. But note what the question does not threaten: the royalty percentage, the rent seniority, the corners, or the payout. A GLP-1 decade would make McDonald's grow slower. It would not make it stop being the landlord.
Nobody fights the landlord on his own corner
| Arena | Who | Where McDonald's stands |
|---|---|---|
| Burgers / QSR | Burger King · Wendy's · Five Guys | ~2× the US sales of the next burger chain (approx.) — scale unchallenged |
| Chicken (the growth war) | Chick-fil-A · Popeyes · Raising Cane's | The real fight — chicken hit beef parity (~$25B); McCrispy $1B+ |
| Value / trade-down | Grocery & home cooking | The 2024–25 battle; value reset won round one |
| Coffee & beverages | Starbucks · Dutch Bros | CosMc's experiment killed in 18 months — learnings absorbed, discipline shown |
| Delivery economics | DoorDash · Uber Direct | Orders flow through MCD's own app — data kept, commissions contained |
The pattern: nobody competes with McDonald's on its actual business. Rivals fight it on menu and price — the tenant layer — while the royalty-and-rent layer has no competitor at all, because no one else owns 45,000 corners with a brand on top. The one competitive war that matters commercially is chicken, where Chick-fil-A earns more per US store than anyone in the industry (approx.) and McDonald's is responding with its biggest menu push in a decade. And note the CosMc's story for what it says about management: a beverage-shop experiment launched in 2023, evaluated honestly, and shut in eighteen months — the anti-empire-building reflex Buffett spends half his letters begging CEOs to develop.
System-first operators · and a board that clawed back $105M
No founder, no family, no control block — this is the professional-manager archetype, which makes the system's discipline the thing to underwrite.
The integrity test, passed the hard way. In 2019 the board fired CEO Steve Easterbrook over a relationship with an employee — then discovered he had lied about others, sued him, and clawed back the entire $105 million severance, one of the largest clawbacks in history. The SEC later fined him and barred him from officer roles. Boards reveal themselves in exactly these moments; this one chose the shareholders over the club. Capital allocation is the simplest on our board: FY2025 returned ~$5.1B of dividends plus ~$2.1B of buybacks — essentially 100% of the $7.2B of free cash flow — a pattern running for decades. Shares are down ~16.5% over ten years (854M → 713M); the dividend has been raised every single year since 1976 — 49 consecutive increases, with the 50th due in October, which would crown it a Dividend King. The one deliberate eccentricity: it is all financed against the real estate, which brings us to the balance sheet.
A flat top line hiding a compounding machine — and one deliberate shock
| Metric | Value | Read |
|---|---|---|
| Revenue (FY2016 → FY2025) | $24.6B → $26.9B | ◆ ~flat — BY DESIGN (refranchising swaps revenue for margin) |
| EPS diluted (FY2016 → FY2025) | $5.44 → $11.95 | ▲ ~9%/yr — the per-share machine under the flat top line |
| Operating margin (decade) | 31% → 46% | ▲ landlord economics fully expressed |
| Net margin (TTM) | 31.6% | ▲ a third of every reported dollar |
| ROIC · ROCE (TTM) | 17.4% · 22.6% | ▲ excellent for an asset-heavy royalty |
| Free cash flow (FY2025) | $7.2B | ▲ dividends + buybacks ≈ 100% of it |
| Net debt / EBITDA · interest cover | 3.6× · 7.9× | ◆ real leverage — serviced by rent |
| Shareholders' equity | −$1.8B | ◆ negative ON PURPOSE — see below |
| Shares outstanding (decade) | 854M → 713M | ▲ −16.5% |
| Dividend (TTM · streak) | $7.44 · 49 raises | ▲ yield 2.8% · payout ~60% · 50th raise due Oct |
Now the deliberate shock, so it never surprises you: McDonald's shareholders' equity is negative — minus $1.8 billion against $54.8 billion of debt. On a screen, that looks like distress. In reality it is arithmetic, not agony: decades of buybacks and dividends have returned more than the company ever retained, and the real estate securing the debt sits on the books at cost — land bought in 1965 carried at 1965 prices, worth perhaps $120 billion at market (approx.). The bond market understands: it lends to this royalty stream at investment-grade rates as if it were infrastructure, because it is. The honest caveat stands regardless: 3.6× net debt to EBITDA means the margin of safety lives in the durability of the rent, not in the printed balance sheet — a structural royalty decline (the GLP-1 scenario) would make today's leverage look different. That is the trade this company has knowingly run for fifty years, and so far the rent has always arrived.
The first stock on our board priced below its own DCF
| Yardstick | Today | Forward | Read |
|---|---|---|---|
| P/E — reported earnings | ~22x | ~20x (FY26) · ~17x (FY28) · ~15x (FY30) | below its own recent history (approx.) |
| P / Free cash flow | ~27x | — | fair for bond-like cash |
| EV / EBITDA | 16.2x | — | the debt is in this number — still reasonable |
| Dividend yield | 2.8% | 50th consecutive raise due October | near the highest of the past decade (approx.) |
| Price vs 52-week range | at a fresh low | −22% from the $342 high | Mr. Market is bored of the landlord |
Savour this section, because after fifteen analyses it is a first. The automated DCF — the model we have taught you to distrust on every AI-capex giant — says McDonald's is worth $310.83, seventeen percent ABOVE the price. And this time the model deserves your attention: a DCF is precisely the right tool for a royalty-and-rent stream with fifty years of history, modest reinvestment needs and no construction-year distortions. The analysts agree from the other direction: mean target $343.50 (+30%), and — uniquely on our board — the lowest target on the street ($305) sits fifteen percent above the market price. Every published number, model and human, says this is worth more than $265. What does the market know? Mostly a mood: GLP-1 headlines, a soft low-income consumer, and the great rotation of 2026 money toward anything with "AI" in the filing. At ~20× forward for a 9%-a-year per-share compounder with a 2.8% yield about to notch its fiftieth consecutive raise, the margin of safety is modest but genuine — and it was bought for you by boredom, the cheapest discount there is. The zone deepens below ~$250, where the yield crosses 3%.
The quietest docket of our July batch — verified July 2026
Verified the week of publication — and the contrast with this batch's siblings is itself the finding: Tesla carries ~21 litigation tracks and a regulator one step from recall power; Meta faces a $1.4 trillion demand in August; McDonald's docket is ordinary corporate weather. The 2024 E. coli outbreak (slivered onions; 104+ sickened, one death) produced individual injury suits now moving toward settlement and a proposed class action still in early stages — management spent ~$100M making franchisees whole and traffic recovered within two quarters (approx.). The franchisee tech-fee dispute (~$70M) sits in active mediation, part of a permanent negotiation that flared in late 2025 over ~$170M of new fees — watch it, because the moat's execution layer is 2,000+ owner-operators, but the National Owners Association has been fighting corporate since 2018 without ever damaging the system's economics. Assorted class actions (wages, no-poach claims, franchising discrimination) follow the company's long-standing pattern: settle, adjust, move on. The genuine risks are the secular ones priced throughout this report — the molecule, the squeezed customer, and leverage that assumes the rent keeps arriving.
I have made expensive mistakes in every decade of my career, but few that I confessed as plainly as this one: in 1998 I told Berkshire's shareholders that "my decision to sell McDonald's was a very big mistake — overall, you would have been better off last year if I had regularly snuck off to the movies during market hours." We had owned thirty million shares. I sold them because the business seemed to require more re-decision from its customers than a Coca-Cola — people choose where to eat every day, I reasoned, and habit is thinner protection at a lunch counter than in a bottle. The three decades since have graded my reasoning: the stake I sold would be worth some ten billion dollars, the dividend has been raised every year without exception, and the company I mistook for a restaurant chain kept revealing itself as what its first president always said it was — a real-estate business that happens to sell hamburgers.
Let me describe what I gave up, precisely. Ninety-five percent of McDonald's restaurants are owned by other people — small businessmen whose life savings sit in the kitchen and who bear every cost that makes restaurants hard: the labor, the beef, the health inspections, the $20 minimum wages. McDonald's Corporation stands above that struggle collecting a royalty on their sales and — the masterstroke — rent, on corners it bought decades ago and carries at prices from another century, collected senior to everything else, enforceable by the lease itself. That is how a hamburger company reports forty-six percent operating margins; that is how earnings per share doubled over a decade in which reported revenue went nowhere; and that is why the balance sheet's negative equity — which frightens every screening tool ever built — is merely the accounting shadow of fifty years of giving shareholders back more than the business ever needed to keep.
The rubs are real and I will not shrink them. The core customer is squeezed, and the value pricing that won back the traffic comes partly out of the franchisees' pockets — the system's permanent internal negotiation, currently in a mediation room over technology fees. The leverage is genuine: three-and-a-half turns of debt against the cash flow, safe exactly as long as the rent keeps arriving. And there is the molecule: appetite-suppressing drugs may reach thirty million Americans by decade's end, and no honest analyst can tell you today what that does to a business built on appetite. My own judgment is that it shaves the growth rather than breaks the model — menus have outlasted every dietary revolution since I've been alive, and rent does not care what the tenant sells — but it is the reason this report scores growth a six and not an eight, and it belongs on your quarterly checklist, not in your blind spot.
Now the price, and the small miracle in it. For fifteen analyses on this shelf I have taught you to distrust the discounted-cash-flow model — it slandered Alphabet, defamed Amazon, and only told the truth about Tesla by accident. Here, for the first time, the model is in its natural habitat: a fifty-year royalty stream with modest reinvestment and no construction-year distortions — and it reads $311 against a $265 price. The analysts' average says $343; even the most bearish published target on Wall Street, $305, sits above the market. Every arithmetic voice says the same thing, and the market disagrees out of nothing more substantial than boredom — the stock sits at a fresh fifty-two-week low because 2026's money wants artificial intelligence and McDonald's merely offers compounding. Twenty times forward earnings, a 2.8% yield weeks from its fiftieth consecutive raise, for the second-most knowable business we have ever scored. I have paid more, for less, with greater confidence than the facts deserved.
So the decision writes itself, and this time I will not repeat 1998: buy the landlord. Buy it for what it actually is — not a bet on burgers but a claim on royalties and rent from forty-five thousand corners, run by managers whose board clawed back a hundred and five million dollars from its own CEO on principle, returning essentially every free dollar to you. Start here at the fifty-two-week low; add below $250, where the yield crosses three percent; reinvest the dividends and let the fiftieth raise compound into the sixtieth. Check the franchisee mood and the GLP-1 data twice a year, and otherwise do what I failed to do thirty years ago — nothing. Some mistakes you only get to correct once the price comes back. It has.