Are AI agents actually disintermediating online travel platforms?

Mauricio Prieto, writing in Travel Tech Essentialist, argues that seven months after one of Skift's 2026 Megatrends predicted "cracks forming in the OTA oligopoly" and warned it would be "extremely difficult to see that the OTAs would emerge from the LLM platform shift unscathed," the actual stock performance of the major online travel agencies has not borne that out. Expedia hit a new all-time high of $327 on August 5, the day of its Q2 report, after revenue rose 14% and gross bookings grew 12%, and closed at $311 on August 7 — up 61% over the trailing 12 months. Airbnb had "its biggest single-day stock rally ever" on August 7 after raising its annual revenue forecast for the second time that year, closing at $178, up 17% that day and 51% over 12 months. Booking Holdings "has gained back most of what it lost earlier this year to that same disintermediation narrative."

Prieto's reasoning is that AI may commoditize discovery — the search-and-compare step — without disintermediating the companies that hold "supply, payments, service, trust, and the ability to complete a booking." Those functions, he argues, "may become more valuable infrastructure" precisely because an agent still needs a platform to execute and guarantee the transaction. He qualifies the finding rather than closing the question: "Seven months is not seven years," and the pattern is uneven across the sector — Trip.com, eDreams ODIGEO, MakeMyTrip and Tripadvisor are all down over the past 12 months, while Trivago (a metasearch business) and the big three OTAs outperform. He concludes that "'OTA' is too broad a category for the AI-disintermediation story," since different platforms carry different economics, geographies, and exposure to the agentic shift.

The finding sits in tension with the mechanism described in Why do consumers delay adopting agentic search?, which models reasons a platform might rationally resist or delay agentic search even where it benefits consumers — Prieto's stock data is an early, concrete data point suggesting the incumbent travel platforms are not (yet) losing value to that shift. It also complicates Will brands succeed by building their own AI shopping agents?: Accenture's framing treats disintermediation as a binary threat that brands must out-compete, while Prieto's evidence points to fulfillment infrastructure — supply, payments, trust — as a chokepoint that can hold value even as discovery itself commoditizes. The companion claim in Does caching fares give infrastructure providers pricing power?, drawn from the same piece, offers one account of where that infrastructure value is concentrating.

A single essayist's reading of roughly seven months of share prices for a handful of named tickers is circumstantial market evidence, not a structural or causal test of whether agents disintermediate travel platforms — it is confounded by overall travel demand, earnings beats, and macro conditions that have nothing to do with agents, and Prieto himself flags the time horizon as too short to settle the question. The implication, held no more strongly than the evidence allows, is that functions beyond discovery — fulfillment, payment, trust — may be where value concentrates as agentic commerce grows, a hypothesis this data is consistent with but does not prove.