AI in Travel: Threat or opportunity?

AI is emerging as a structural threat to online travel agencies, with suppliers like Accor, Hyatt, and Wyndham pushing direct acquisition through platforms such as ChatGPT to reduce OTA commissions. Bernstein says the key variable is the AI revenue model: ad-supported systems may preserve OTA take rates, while subscription or user-funded models could enable a 0% take-rate environment and disintermediate Booking and Expedia. The near-term impact is more sentiment-driven than fundamental, but the long-term competitive pressure on travel distribution is clearly negative for OTAs.

Analysis

The market is underestimating how quickly AI distribution can compress the OTA toll bridge even before a fully autonomous booking workflow exists. The first-order threat is margin pressure, but the second-order threat is that suppliers gain leverage to negotiate lower commission structures across all channels once a credible direct-to-AI path exists. That matters more for BKNG than EXPE because the former’s scale advantage is partly built on being the default search-and-conversion layer; if discovery migrates into AI interfaces, that default status becomes less defensible.

The near-term setup is not a sudden revenue cliff but a slow bleed in take-rate economics and customer acquisition efficiency. The biggest risk to the bears is that AI platforms initially monetize like search, not like a pure subscription utility, which would preserve paid placement and allow OTAs to simply re-bid their way into visibility. In that regime, the winners are the AI gatekeepers and the hotel tech middleware layer, while hotel brands gain bargaining power without necessarily having to fully disintermediate the OTAs.

The consensus is likely too focused on trust and too little on habit formation. Trust gaps in consumer behavior can close faster than expected once AI becomes embedded in the booking funnel through adjacent use cases, and the transition may start with research and meta-search before ending in full conversion. The real dislocation would be a consumer-funded AI model that removes the need for supplier commissions altogether; that is a multi-quarter to multi-year risk, but it is exactly the sort of structural shift that deserves a valuation discount today.

For now, the tradeable catalyst path is sequencing: first AI discovery, then AI-assisted booking, then payment. Each step increases the probability of lower OTA take rates and higher supplier bargaining power, and each step can happen without dramatic user adoption headlines. That makes this a slow-burn negative for OTAs with asymmetric downside if AI interfaces capture even a modest share of travel intent.

Key Decisions for Investors

• Fade strength in BKNG and EXPE into rallies: initiate 3-6 month put spreads or call spreads on the downside, targeting a modest re-rating compression if AI distribution headlines keep recurring; risk/reward favors defined-risk bearish exposure rather than outright shorts given near-term resilience.
• Pair trade: long hotel/software enablement names tied to channel management and inventory connectivity vs short BKNG/EXPE for a 6-12 month horizon; the spread works if AI adoption shifts value from conversion monopolies to plumbing and orchestration.
• Buy BKNG downside protection around earnings or major AI-platform travel announcements: 1-2 quarter tenor puts financed by selling far OTM puts or calls to express a slow erosion thesis while limiting carry bleed.
• If you want to stay directionally bearish but avoid factor risk, short EXPE against a basket of supplier names that benefit from direct distribution leverage; EXPE is more exposed to a weaker trust moat and has less scale to absorb channel compression.