Travelers now use generative AI tools to plan and book trips roughly two years ahead of the suppliers, intermediaries and platforms that sell to them, according to new research flagged by trade publication Hospitality Net.

That gap shapes how demand forms, how product gets filtered and which commission-bearing inventory reaches consumers. When buyers run queries on ChatGPT, Google Gemini, Microsoft Copilot or Anthropic's Claude before they see an airline, hotel or tour-operator website, the catalog a seller controls risks dropping out of the consideration set entirely.

The headline finding reverses the usual technology sequence. Suppliers have spent 24 months racing to embed AI into chatbots, revenue management and content pipelines. Operators including Booking Holdings, Expedia Group, Airbnb, Skyscanner, Marriott, Hilton, IHG and Amadeus have shipped or piloted large language model features. The same research indicates the buying public moved faster.

What is driving consumer adoption?

Three forces account for the speed of demand-side uptake:

• Free or low-cost chatbot access collapsed the experimentation cost for trip planning.
• Model upgrades that handle multi-step itineraries removed the need for agent expertise.
• Social distribution of prompt templates lowered the learning curve for non-technical travelers.

Where does the two-year gap hit revenue?

For sellers of travel, the revenue line that matters is whether AI-mediated shopping routes through their booking engines, or detours around them.

If a conversational assistant surfaces a hotel directly, the property keeps the conversion but bypasses the OTA commission. If the assistant surfaces a metasearch link, the merchant pays an auction-priced referral. If it surfaces nothing attributable, the booking migrates to whichever supplier has the strongest direct footprint or the most licensed data in the model's training corpus.

Search-engine and AI-platform partnerships reshape who benefits. As large language models integrate travel content directly into chat responses, the LLM layer becomes a new customer acquisition channel. Distribution chiefs at chains and bed banks will need to monitor referral traffic from these sources the same way they watch Google hotel-module share today.

Who loses share first?

Smaller operators face the steeper risk. An independent hotel, a regional tour operator or a niche destination marketing organization lacks the staff to negotiate data deals, publish structured feeds or audit how a model represents their product. A two-year demand-side gap compounds into a multi-year supply-side disadvantage once models lock in their citation patterns.

DMO budgets will come under pressure too. Tourist boards have spent years optimizing content for Google's "things to do" panels. Optimizing for generative answers — and tracking when a destination is named, cited or ignored — requires a different measurement framework.

The two-layer stack

AI has split the distribution stack into two layers: the model layer where intent now forms, and the merchant layer where transactions settle. Sellers that treat the model layer as marketing collateral rather than a primary storefront will see their share of voice erode before their conversion rates fall.

The consumer pace reported by Hospitality Net points to a decisive 12-month window. Suppliers that have not wired their inventory, content and pricing into machine-readable formats risk watching competitors capture the queries first.