Believe it or not, flying is cheaper than it has been in years. Now, many fear that AI could disrupt this trend.
Nominal airfares—not adjusted for inflation—are 12% lower than they were in February 2020. Meanwhile, the price of all items in the consumer price index has gone up 24% during that time—a gap of 36 percentage points. Plus, inflation-adjusted airfares are 41% cheaper than 10 years ago. And June 2025 was the second-cheapest month ever for inflation-adjusted airfares.
[time-brightcove not-tgx=”true”]
It’s against this backdrop that we learned Delta Air Lines has been using AI to help set the price of some tickets.
“A super analyst” working “24 hours a day, 7 days a week” to optimize airfares is how Delta president Glen Hauenstein described its AI initiative to investors. In late 2024, Delta was using AI on about 1% of fares; by the end of 2025 it’s aiming for 20%.
The most common reaction has been outrage. Democratic Arizona Senator Ruben Gallego—a potential 2028 presidential contender—encapsulated this view, calling it “predatory pricing” and vowing he “won’t let them get away with this.”
Dynamic airfare pricing isn’t new. It’s as old as the modern aviation industry. And though dynamic pricing super-powered by AI could introduce new confusion and complications, it could help lower flight costs further.
The history of dynamic airfare pricing
Prior to the late 1970s, airfare had a mandated price: typically 6 cents per mile flown. But after deregulation, airlines were permitted to charge whatever they wanted, and flight prices plummeted. Today, airfare costs half of what it did in the 1970s, making travel a reality for much of society—over half of Americans now fly every year—rather than a hobby exclusively for the wealthy.
Fifty years ago, flight prices were one-size-fits-all. Thirty years ago, flight prices changed by the week. A decade ago, flight prices changed by the day. Today, airfare is near all-time lows, yet it’s the most volatile purchase that consumers regularly make. As airfare shifted from standardized to dynamic pricing, prices went down but confusion went up.
I remember watching a flight once from Atlanta to Amsterdam. When I searched on Monday, prices were $800 roundtrip. When I searched again on Tuesday, prices had dropped to $300 roundtrip. By Wednesday, prices were up to $1,300 roundtrip. This is for the exact same flight on the exact same travel dates. The idea of a standard “what it costs to fly” somewhere is a thing of the past.
Next time you’re on a flight, ask the person sitting next to you what they paid. You won’t be surprised to hear a very different price than you paid, even as the two of you are having the exact same travel experience. On a flight with 200 passengers, it wouldn’t be unusual for them to have paid 150 different prices.
Tensions turbocharged by AI
Do people like the confusion of dynamic flight prices? Of course not. Do people like cheap flights? Yes. And that is the inherent tension of airfare today—a tension that AI could turbocharge.
The remaining question is this: Will AI push Delta’s fares up or down?
The price of some flights are likely to go up as a result of AI. However, I am skeptical that this will be the case for all—or even most—fares. Airlines can’t just increase prices at their leisure for the same reason McDonald’s can’t charge $30 for a chicken sandwich; customers will flock to their competitors. If Delta’s charging $200 more than United for a flight to Chicago, who will book Delta?
Still, without transparency and regulation into its AI-powered pricing model, Delta holds the cards here and is incentivized to prioritize maximizing revenue. If AI wasn’t generating more revenue, Delta wouldn’t be using it. And as Hauenstein noted to investors, its AI foray has been producing “amazingly favorable” results.’
That gets at a subtle but important distinction here, though: generating more revenue isn’t always the same as charging higher fares across the board.
The primary reason flight prices have fallen so sharply in the past decade is that they don’t matter for airlines’ bottom lines nearly as much as they used to. Fifty years ago, the majority of airline revenue came from economy airfare. Today, the majority of airline revenue comes from sources other than economy airfare. It comes from premium seats, credit cards and frequent flyer miles, corporate contracts, cargo, add-on fees, and other travel commissions.
Between 2011 and 2018, Delta’s economy ticket revenue went down by $1 billion, but their overall revenue grew by $9 billion.
And this is not unique to Delta. In 2024—a year in which American Airlines generated $846 million in profits—the airline spent 17.6 cents per seat mile flying planes but brought in just 16.9 cents per seat mile in passenger revenue. Their flights lost money and they still turned a profit thanks in large part to billions in revenue brought in by the airline’s AAdvantage loyalty program.
Read More: How to Save OpenAI’s Nonprofit Soul, According to a Former OpenAI Employee
Think back to the last time you bought a flight. Here’s an incomplete list of all the upsell attempts the airline likely made: cabin upgrade, seat selection, checked baggage, hotels, car rentals, tours, the airline’s credit card, additional frequent flyer miles. You may have opted for none of these items, but on average, most travelers will get at least one or two.
The airline business model is not unlike a restaurant’s, which sells some items at cost (or even at a loss) and others with high markups. For airlines, economy seats are like a steak, sold at virtually no profit. Credit cards and premium seats, meanwhile, are the sodas and alcohol, sold way above cost. Much like how pricey alcohol allows restaurants to serve steak, credit cards and premium cabins are what allow airlines to generate record-high profits—even during periods of record-low economy fares.
So while fears of AI driving up airfare are rampant, on average I would expect the opposite to happen. It’s not that Delta wouldn’t love to make an extra $100; it’s that they’re agnostic whether that $100 comes from ticket revenue, credit card sales, hotel commissions, or any other source.
Within the United States, air travel is relatively competitive—on Aug. 11 for instance, there are 47 nonstop flights between NYC and LA alone—and that competition between airlines is the single biggest driver of cheap flights. But once you’ve bought a Delta flight, they’re the only ones who can sell you things like seats and checked luggage. And they’re the first to know your trip is official; how about a hotel deal or car discount to round out your vacation? In other words, it’s critical for airlines to make the sale, and the earlier the better so they have more time to market add-ons. What’s the most reliable way to sell flights? Make them cheaper.
I don’t begrudge people’s fears when they hear about an airline using AI to help price flights. The airlines certainly haven’t earned any benefit of the doubt when it comes to pricing transparency. But Delta’s new AI pricing initiative doesn’t forbode the end of our ongoing Golden Age of Cheap Flights. On the contrary, as airlines continue to diversify their revenue streams, I’m hopeful we’ll see AI drive fares down even more.
For decades, flight prices were simple and expensive. Today, they’re confusing and cheap. There’s nothing wrong with wishing for simple and cheap, but I’m grateful to live in a time when air travel is—for now—still affordable for the masses.