Divergent Compute.AI Economic Think Tank

Markets / Industries / Travel & Hospitality

Industry view · Travel & Hospitality

The booking funnel is collapsing into a prompt.

Travel is an $11.6 trillion slice of the world economy, and for two decades the online travel agencies owned the path from dreaming to booking. AI is rewriting that path — discovery now starts inside a chatbot, itineraries assemble themselves, and agents threaten to book around the middlemen entirely. The question every player is now asking: does the agent route through me, or around me?

$11.6T
Travel & tourism's contribution to global GDP, 2025
WTTC
$1.67T
Gross travel bookings, 2025 — $1.07T of it online
Phocuswright
30%
US travelers now using AI extensively to plan — double a year earlier
Skift 2025
~$1.2→8.4B
Travel & hospitality AI market, 2026 to 2030 estimate
IndustryArc

01 · The thesis

From a funnel to a spiral

The old journey was a funnel: a traveler browsed, compared across tabs, and converted on an OTA that had bought the keyword. Generative AI compresses that into a single exchange — one prompt, one shortlist, one booking — and the industry has started calling it a spiral: no awareness phase, no slow drip, just intent and action. Skift and McKinsey find 90% of major travel companies have launched a generative-AI project and over half are testing agentic capabilities.

That collapses the value chain unevenly. Discovery and planning are moving fastest, into assistants travelers already use. Booking — where the money is — is the contested layer: if an AI agent can complete the reservation, the OTA's two-decade grip on conversion is in play. The visible-but-unsettled frontier is agentic booking, and where it lands decides who keeps the margin.

Pace of AI disruption by stage — Divergent Compute assessment

02 · Public players & exposure

Who routes through, who gets routed around

The same forces help and threaten the incumbents. We plot the listed players on two editorial axes — how exposed each is to being disintermediated by AI agents, against how ready its data, brand and inventory are to be the agent's answer. The numbers in the table are sourced; the placement is our read.

Positioning — editorial assessment, not a sourced metric. Bubble = approximate relative scale.
CompanyStanceThe sourced fact
The map is Divergent Compute's editorial positioning, offered as a lens, not a measurement. Every figure in the right-hand column is drawn from a company filing, earnings call, or named report — see Sources.

03 · The two clocks

The spend is real. The payoff is starting to show.

Divergent Compute reads every industry as two clocks — how fast the capital goes in, and how fast the return comes out. In travel the second clock is, unusually, already ticking: the early wins are in cost-to-serve, not just top line.

Travel & hospitality AI market, 2026 estimate vs 2030 forecast (IndustryArc).

Booking Holdings pulled $250M of operating-expense savings out of a 2025 transformation program and expects a larger run-rate in 2026, while its "connected trip" bookings grew in the high-20% range.

Airbnb says its AI assistant now resolves over 40% of guest service issues and has cut cost per booking about 10% year-over-year.

Expedia handles 30%+ of self-serve support interactions with AI and is feeding 70 petabytes of trip history into proactive, "self-healing" itineraries. The payoff clock here leads most industries we cover.

04 · Private flagships

The AI-native challengers

Seven of the ten largest AI startup deals in online travel in 2025 were customer-service plays — but the louder bet is the AI-native agent that plans and books. The standouts, with disclosed funding:

05 · Signals

What moved, and what to watch

06 · The exposure read

Who's defensible, who's at risk

AI agents reward clean, machine-readable supply and punish friction. The line between winner and loser runs through who owns inventory, brand and behavioral data — and who is merely a tollbooth the agent can route around.

Defensible

  • Decades of behavioral data + payments rails. Booking's connectivity to millions of properties is exactly what an agent needs as a partner.
  • Exclusive inventory & strong consumer brands. Hotels and platforms with supply you can't get elsewhere stay in the answer.
  • Clean, structured, API-ready inventory. IHG is rebuilding its hotel data "from links to answers" so agents can read it.
  • High-touch, luxury and experiential. Where judgment and access matter, human-plus-AI keeps pricing power.

At risk

  • Pure metasearch & lead-gen. If the agent is the comparison layer, the comparison business gets squeezed.
  • Legacy TMCs & manual revenue management. Automatable workflows are the first to be absorbed.
  • Messy inventory invisible to AI search. Properties without clean rates and APIs get deprioritized before a traveler ever appears.
  • Thin-margin intermediaries. Any tollbooth an agent can route around is a target.
Even the disintermediation case has limits: in March 2026 OpenAI pulled back from direct checkout inside ChatGPT, sending Expedia up ~13% and Booking ~8% — a reminder that booking is harder to automate than discovery, and the incumbents are not passive.

Sources

Where this comes from