How Google Makes Money on a Trip It Never Sells

I plan trips the slow way. A fare calendar open in one tab, a hotel award chart in another, a map I keep zooming into for no reason except that I like looking at it. I'll spend an hour finding a connection that saves ninety dollars and call it a good time. For years I've half-wondered whether that makes me a relic — whether the whole ritual of the hunt is a thing people my age do, and everyone younger just asks the phone.

This year the phone got a lot better at answering. So I went looking for what that does to the business of travel, and found something more interesting than my own obsolescence.

Start with a puzzle. In 2011, Google bought a quiet airfare-engine company called ITA Software for about $700 million, cleared antitrust with a consent decree, and launched Google Flights. The online travel agencies — the Bookings and Expedias of the world, the sites that actually take your booking and your credit card — were supposed to be dead men walking. Google had the traffic, the data, and now the fare engine. Why would anyone start a trip anywhere else?

Fifteen years later, the online travel agencies are the biggest they have ever been.

Booking Holdings just closed its biggest year on record. Expedia grew too. And Booking is now worth about $135 billion — more than Delta, United, and American put together.

The extinction event created two giants. Why it went the other way is the part worth understanding.

Google Flights has never sold a single ticket. Search a route, find a fare, and Google hands you off — to the airline, or to an OTA — to actually buy it. Google didn't want to be the store. It wanted to be the shelf, and to charge for the spot at eye level.

That turns out to be a much better business than being the store. An online travel agency has to contract with hotels, run call centers, eat cancellations and chargebacks, and own the customer when a trip falls apart at 11 p.m. in a foreign airport. Google skips all of it and takes a cut of the demand at the top of the funnel. Why swallow the operational mess when you can rent the shelf to the people who already run it well?

So the OTAs didn't get replaced. They got a landlord, and they paid up.

Booking and Expedia spent roughly $15.5 billion on marketing in 2025, and much of that flows straight back to Google. This isn't new: the two spent over $11 billion on marketing back in 2019, most of it to Google, and at one point Booking was handing Google more than a billion dollars in a single quarter. The companies Google was supposed to kill became its best customers.

They also spent the decade building things Google didn't want to build. Flights are commodities — thin margins, easy to compare. Hotels are the opposite: millions of independent properties, each its own negotiation, and the fat margins that come with all that fragmentation. Google has no appetite for signing up a boutique hotel in Lisbon one property at a time. The OTAs do it for a living. Booking wrapped flights, hotels, cars, and attractions into a single “connected trip,” and the number of its travelers buying more than one piece at once rose 20% in 2025. They built apps and loyalty tiers that bring people back without paying the Google toll every time. They even bought the metasearch layer that competes with Google Flights: Booking owns Kayak, Expedia controls trivago. Over fifteen years, the online travel agencies turned themselves into infrastructure.

Which brings me back to the phone getting better at answering. Through late 2025 and 2026, Google started folding travel into its AI. There's a planning canvas that builds itineraries from live flight and hotel data. A “flight deals” tool that now works in more than 200 countries. And the headline feature: agentic booking, where you describe what you want and the AI arranges it. Restaurants went first, through OpenTable and Resy. Flights and hotels are next.

You would think an AI that books your whole trip is the thing that finally cuts out the middleman. Then you look at who Google named as its booking partners.

Booking.com, Expedia, Marriott, IHG, Choice, Wyndham. And when the AI is ready to book, it hands you a direct link to finish the purchase with one of them. Google isn't taking the payment. It's routing you to the OTA. Again.

As the travel-tech writer Markus Busch put it in Hospitality Net: a company does not launch a plan to disintermediate the middlemen with those exact middlemen as its founding partners. This isn't Google cutting the OTAs out. It's Google adding a floor above them. The traveler now arrives at the decision inside Google's AI instead of on Booking's homepage — and that surface, the place where the choice actually gets made, is what moves to Google. The merchant keeps the booking, the fulfillment, and the liability. Google keeps the customer.

Which answers the question I actually care about: how does AI make money on travel? The same way Google always has. Own the surface where the decision happens, and charge the people who need to reach it. Google Flights sold placement on the shelf. AI Mode sells placement in the answer. The tollbooth doesn't disappear — it moves from the blue link to the sentence the AI writes back to you.

The tell is what Google hasn't said. Its big travel announcement laid out every feature and went silent on how it charges. That silence is the strategy. The auction that made Google Flights a money machine — pay more, appear first — will reassemble inside the AI, whether as sponsored partners or a cut of every handoff. For the OTAs, the danger was never being bypassed. It's paying rent to a landlord who just moved up a floor, owns the front door, and can raise the toll whenever it likes. Booking already sees it. Through its Priceline brand, it has an AI agent named Penny that went fully agentic this June — a booking assistant of its own, built to keep travelers on its side of the wall.

So — am I a relic for enjoying the search? I went into the research half-expecting to be told yes. It says the opposite. The people who stay engaged in planning a trip report more anticipation, more satisfaction with the trip itself, and more satisfaction with life afterward. The people who hand the whole thing off report less. There's a ceiling — plan too much and the benefit fades — but the joy is real, and it's measurable.

It's not about age. It's about engagement, and agentic booking is built for the enormous majority who never wanted to engage — who feel the hunt as a chore and would gladly give it to a machine. That's most travelers, which is why the AI will win the default. But it leaves an opening. The hunt survives as something you opt into: the enthusiast, the points optimizer, the person who likes looking at the map for no reason. My bet is that the tools which preserve the fun of planning, instead of automating it away, turn out to be worth paying for. Which is its own answer to how AI monetizes travel.

I'm not old. I'm just the customer everyone in this business is fighting to own.