The Airline Doesn't Fly

In 2026, the International Air Transport Association projected a record year for the global airline industry. A trillion dollars in revenue. Five billion passengers. Net profit per passenger: seven dollars and ninety cents.¹

Seven dollars. In the best year the industry has ever had. Less than an airport sandwich.

In 1999, Warren Buffett described what the numbers said about the previous nine decades.

"As of 1992, the money that had been made since the dawn of aviation by all of this country's airline companies was zero. Absolutely zero. I like to think that if I'd been at Kitty Hawk in 1903 when Orville Wright took off, I would have been farsighted enough, and public-spirited enough, to shoot him down. I mean, Karl Marx couldn't have done as much damage to capitalists as Orville did."²

He was not being theatrical. He was reading a balance sheet. From the Wright Brothers through 1992, the cumulative net profit of every American airline combined was zero. The industry that connected continents had not produced a single dollar of net wealth for its owners across nine decades.

Between 1945 and 2000, the global airline industry generated cumulative net profits of $36 billion. Fifty-five years of operations at a combined margin of 0.8%. In the following decade, the industry lost $49 billion, a third more than it had earned in the previous half-century combined.³ Then COVID eliminated another $201 billion in three years.⁴

Then came the most recent recovery. Consolidation reduced the US market from nine major carriers to four. Capacity discipline and fuel hedging produced the first sustained profitable stretch in the industry's history. Airlines finally earned their cost of capital. McKinsey called it "a remarkable feat."⁵

The cycle is the defining feature. Profits accumulate slowly during years of discipline, then vanish in a single shock. High fixed costs. Perishable inventory. Price-sensitive customers. A product that becomes worthless the moment the gate closes. Every textbook explanation of why flying is a bad business is correct.

What the textbooks do not explain is why the industry, after a century of failing to produce consistent returns from transportation, is more valuable than it has ever been.

Buffett himself tested his own conclusion. In 2016, after decades of calling airlines structurally flawed, Berkshire Hathaway bought significant stakes in all four major US carriers. Delta, American, Southwest, United. The thesis had apparently changed. In April 2020, he sold the entire position at a loss of roughly $5 billion. "The world has changed for the airlines," he told shareholders. "I don't know how it's changed."⁶

He was right about the confusion. The world had changed. But it had not changed in the direction he imagined. The airline industry had not become a better transportation business. It had quietly become a different business altogether.

What the balance sheet said

On November 29, 2011, AMR Corporation, the parent of American Airlines, filed for Chapter 11 bankruptcy protection. The company reported $24.7 billion in assets and $29.6 billion in liabilities. Equity was negative $7.1 billion. The stock was delisted.⁷

The most valuable asset in the company was not visible on the balance sheet in any meaningful way. AAdvantage, the frequent flyer program, existed as part of the airline's operations. It had no separate financial statements. No standalone valuation. When AMR's equity was negative $7.1 billion, the program that would later be appraised at $18 to $30 billion was hiding in plain sight.

What mattered, for understanding what airlines actually are, was what happened nine years later.

In the spring of 2020, COVID grounded most of the fleet. Revenue fell more than 60%. American needed billions in emergency financing. Under the CARES Act, the US Treasury offered loans but required substantial collateral. American offered its most valuable asset. Not the 900 aircraft. Not the gate leases at 350 airports. Not the route authority accumulated over decades.

The AAdvantage frequent flyer program.

At a Wolfe Research conference in May 2020, CFO Derek Kerr stated the valuation directly: "We have an appraised value of that program anywhere from $18 billion to $30 billion." Third-party appraisals placed it between $19.5 billion and $31.5 billion.⁸ At the time, American Airlines' total market capitalization was under $10 billion.

The loyalty program was worth two to three times the airline.

American was not an anomaly. The same year exposed the same inversion at every major carrier.

United Airlines pledged its MileagePlus program as collateral for $6.8 billion in financing. The loan documentation, filed with the SEC, contained financial disclosures the airline had never previously made public. MileagePlus generated $5.3 billion in revenue in 2019 and $1.8 billion in EBITDA. United valued the program at $21.9 billion. The airline's market cap was $10 to $12 billion. MileagePlus was worth roughly twice the airline. United was burning $40 million in cash per day.⁹

Delta raised $9 billion in debt backed by SkyMiles, upsized from an initial $6.5 billion offering after investor demand exceeded expectations. On Point Loyalty later valued the SkyMiles program at $28 billion.¹⁰

Combined, the three largest US airlines raised more than $20 billion in emergency financing during the pandemic. The collateral was not aircraft, ground infrastructure, or route authority. It was databases of customer spending behavior linked to a synthetic currency.

Aviation Week described the shift in 2023: "Historically, aircraft, gates, and other assets had been used as collateral. But it was the loyalty programs that became the ultimate source of financing." On Point Loyalty was more direct: "Once the best kept secret of the airline industry, loyalty programs have irreversibly stepped onto the centre stage of global airline economics."¹¹

The pattern had precedent. In 2005, Air Canada's parent company, fresh from bankruptcy protection, spun off its Aeroplan loyalty program in an IPO that implied a valuation of approximately $2 billion. The program operated independently for twelve years, generating steady profits while the airline weathered fuel shocks, recessions, and competitive pressure. In January 2019, Air Canada repurchased Aeroplan for $450 million in cash plus the assumption of $1.9 billion in unredeemed miles. Total deal value: approximately $2.4 billion. The airline paid $2.4 billion to reacquire the asset it had spun off for $2 billion, because the loyalty program was too valuable to let anyone else operate.¹²

The currency

Airlines create miles at zero production cost. A mile is a unit of a proprietary currency that the airline controls completely, not a unit of distance. The airline sets the issuance rate, the redemption value, the expiration rules, and the total supply. No external regulator constrains any of these decisions. A 2024 paper in the Vanderbilt Law Review argued that airlines exercise more control over frequent flyer miles than a central bank exercises over sovereign currency.¹³

The airline sells this currency wholesale to banks.

When American Express pays Delta for SkyMiles, or Chase pays United for MileagePlus points, the bank is purchasing a synthetic currency at a negotiated rate. The blended wholesale price is approximately 1.5 cents per mile. Sign-up bonus miles cost about a penny each. Miles awarded for ongoing card spending cost closer to 1.5 cents. The bank pays this price because the currency drives credit card acquisition and everyday spending. A consumer who earns 2 miles per dollar on groceries with a Delta Amex card is more valuable to both Amex and Delta than a consumer who just flies twice a year.¹⁴

The cost to the airline of fulfilling a mile, when a customer eventually redeems it for a flight, is approximately 0.72 cents. American Airlines disclosed a 52% operating margin on AAdvantage in a 2021 securitization filing with the SEC. Delta reported 39%. United: 34%. The airline sells a mile for roughly 1.5 cents and delivers the seat for roughly half that. The spread is the business.¹⁵

Then there is breakage. Across the industry, 10 to 20 percent of miles issued are never redeemed. The customer earns them, carries them as an account balance, and never converts them to a flight or an upgrade. The airline has already been paid for these miles by the bank. The fulfillment cost is zero. Breakage is pure margin on top of margins that already exceed 50%.

Before 2018, this economic structure was largely invisible. Airlines bundled loyalty revenue inside passenger ticket income. The accounting standard that changed this, ASC 606, took effect in January 2018. It required airlines to treat miles as a separate performance obligation with its own deferred revenue. Loyalty program liabilities appeared as a distinct line item on the balance sheet for the first time.¹⁶

As of 2024, American Airlines carried $10.1 billion in loyalty-related deferred revenue. Delta: $8.7 billion. United: $7.9 billion. The combined figure across three airlines was $26.7 billion in deferred liabilities representing miles sold to banks but not yet redeemed by customers.

The combined co-brand credit card revenue paid by banks to the Big Three in 2024 was $16.4 billion. American's partnerships generated $6.1 billion. Delta's American Express deal generated $7.4 billion. United's Chase relationship generated $2.9 billion.¹⁷ This money came from banks, not passengers. It came from everyday spending at grocery stores and gas stations. Not from selling airplane seats.

The consumer sees none of this arithmetic. They sign up for a Delta Amex card because it offers miles per dollar and free checked bags. They use the card at restaurants, gas stations, and online retailers. Each purchase generates a payment from Amex to Delta. The consumer flies once or twice a year. They spend on the card every day. The daily transactions generate more revenue for the airline than the occasional flight does.

The inversion

In March 2025, Courtney Miller of Visual Approach Analytics published an analysis that asked a question the industry had never formally answered. What happens to airline profitability when you isolate flight operations from loyalty revenue?¹⁸

The methodology was direct. Take each airline's reported operating margin. Remove the loyalty program revenue. Recalculate.

Every airline went negative. Not one major US carrier produced a positive operating margin from flying passengers in 2024. The loyalty programs generated not some of the profit, not most of the profit, but more than 100% of the profit. The flight operations were a net loss partially offset by the financial services business attached to them.

The Economist confirmed the finding in August 2025: "You might expect America's most valuable airline to earn its keep flying passengers. But you would be mistaken. Without the revenue from its loyalty programme, [Delta] would have operated at a loss."²⁰

American Airlines made the arithmetic unavoidable in its fiscal year 2025 results. Total revenue: $54.6 billion. Net income: $111 million. A margin of 0.2%. AAdvantage generated approximately $2 billion in profit at a 53% operating margin. The rest of the airline, the part that actually flies planes, lost approximately $2 billion. The loyalty program was the airline's earnings, not a supplement to them. Everything else was overhead.²¹

Southwest reveals the inversion most starkly. No carrier has historically been more identified with affordable, no-frills flying. Southwest built its brand on low fares, open seating, and operational simplicity. Its Rapid Rewards program contributed 21.1% of total revenue, the highest share among the Big Five. Remove it, and the operating margin falls to negative 19.9%. The airline that defined itself as a low-cost carrier was kept solvent by the credit cards in its customers' wallets.

The pattern was invisible for decades because airlines did not disclose loyalty economics separately. The accounting change in 2018 made the deferred revenue visible. The pandemic securitizations made the standalone valuations visible. Miller's analysis made the margin inversion visible. Each disclosure revealed the same structural fact from a different angle. In September 2024, the US Department of Transportation opened a formal investigation into airline frequent flyer programs to examine whether they are "transparent and fair."¹⁹ The regulator is catching up to the finding.

The flying has never been the profitable part. The visibility is new. The economics are not.

If the flight operations lose money and the loyalty program generates all the profit, then the economic purpose of the flight network is not to transport passengers profitably. The purpose is to maintain the conditions under which banks will continue paying billions for miles. A route that loses money on passenger revenue but generates credit card sign-ups is a functioning customer acquisition channel, not a failing route.

Where the planes fly

In October 2025, Delta CEO Ed Bastian explained why the airline was expanding service to Austin and Raleigh. The reason was credit card acquisition, not passenger demand. "These are places where we acquire a lot of cards," Bastian said.²²

The route map is a customer acquisition strategy.

This is an operational statement from the CEO of the most profitable airline in the United States, not an inference from financial data. Delta does not add flights to Austin primarily because passengers want to fly there. Delta adds flights because people who live in Austin will sign up for an American Express co-branded credit card when Delta is present in their market. Those cardholders then spend money on groceries, gas, restaurants, and retail. Each dollar they spend earns SkyMiles. Each mile earned generates a payment from American Express to Delta. The flight is the acquisition channel. The credit card is the product.

One-third of all active SkyMiles members now hold a co-branded Amex card. The annual revenue from that single partnership tells the full story.

In the fourth quarter of 2024 alone, American Express paid Delta approximately $2 billion. That single quarterly payment from one credit card company was larger than many airlines' total annual profit.

Delta's diversified revenue, which includes premium seating alongside the Amex income, accounted for 57% of total revenue in 2024. The flight is one event inside a broader economic relationship between the passenger, the credit card, and the airline, no longer the primary revenue event.²²

American Airlines reached the same structural conclusion from a different direction. In December 2024, American signed a ten-year exclusive partnership with Citi, replacing Barclays as its primary co-brand partner. The filing with the SEC projected co-brand revenue growing from $5.6 billion toward $10 billion annually. At full scale, American estimated the incremental pre-tax income from the Citi deal at $1.5 billion per year. That is $1.5 billion in annual profit from a credit card agreement. Not from flying more routes, selling more tickets, or filling more seats.²³

The programs began as a competitive tool. American Airlines created AAdvantage in May 1981 to prevent business travelers from switching carriers. United followed within a week. Every major airline matched within months. The flight was the product. The miles were the incentive to stay loyal.

Forty-four years later, the structure has inverted. The miles are the product. The flight is the incentive that keeps the cardholder engaged.

The pattern

Every system I've examined operates on a belief that nobody checks.

In Formula 1, the belief was about what the product is. Everyone watched the racing. For forty years, the sport was operated as a racing series. The racing was the content that generated an audience, which was the asset that supported a franchise model. When Liberty Media challenged the premise, the result was $20 billion in realized value that had been constrained by the old belief.

In retail, the belief is about whether the economics work. Founders hold a gross margin number that ignores eleven categories of cost between the factory and the bank account. The belief persists because the purchase order arrives before anyone runs the real math.

In the Premier League, the belief is about what determines the outcome. The table looks like a competition decided by tactical decisions and individual talent. Wage spending explains 92% of the variation in league position across twenty years.

In pricing, the belief is about when the economics get decided. Founders think they set the price at the buyer meeting. The price was determined when the manufacturing cost was committed.

In aviation, the belief is about what the airline sells. Every passenger who boards a plane believes they are the customer. They are not the customer. They are the raw material. The customer is the bank that pays billions for the spending behavior the passenger generates. The flight is the cost of acquiring and retaining a credit card holder.

Same structure across every domain. An inherited belief. Decisions that follow logically from it. A structural reality that runs counter to it. And a long stretch where nobody checks, because the system is functional enough that the question never seems urgent.

Implication

The airline industry spent ninety years trying to make money flying people places. It never consistently succeeded. Buffett saw it in 1999 and concluded the industry was a permanently bad investment. He was right about the transportation business. He missed the financial services business growing inside it.

Loyalty programs began in 1981 as a marketing tool. They were designed to fill seats. They succeeded. They also built something nobody planned: a financial services business with 50% margins, growing faster than the transportation business it was attached to, worth more than the airline by almost every measure, and generating revenue whether the passenger flew or not.

Every reader has been inside this system. They have earned miles, carried co-branded credit cards, checked point balances before booking a trip. They have experienced the system while holding a belief about it that the balance sheet does not support.

The airline doesn't fly. The planes are still in the air, but flying was never the business.