Travelers are turning to AI to plan trips — but hallucinations and trust gaps remain

91% of global travelers use AI travel planners, per Klook's survey of 11,000 users, while a Booking.com report found 91% of respondents have concerns and only 35% fully trust AI outputs. The article flags accuracy risks ('hallucinations')—including a ChatGPT route error that turned a 10-minute transfer into 45 minutes—and warns smaller, older or developing-market properties may be disadvantaged by limited digital presence. Booking.com and others are rolling out AI (including OpenAI integrations), and industry experts say structuring data via APIs could reduce errors; adoption is expected to grow despite current limitations.

Analysis

AI travel planners are not a neutral tool — they rewire demand discovery. Because current LLMs lean on well-indexed, high-quality web signals and curated APIs, expect referral concentration to intensify: within 12–24 months the top 10–20% of destinations/properties that expose clean APIs and structured metadata could capture a disproportionate 50–70% of AI-driven bookings, amplifying winner-take-most economics in distribution and advertising spend.

That shift creates a two-layer compute and data prize. On the compute side, personalized multimodal itinerary generation and real-time routing inflate inference workloads materially — we model a 3–5x increase in travel-specific inference per user as planners move from static lists to interactive itineraries over 6–18 months, favoring GPU/infra providers and cloud vendors with fine-tuning and retrieval services. On the data side, tourism boards, OTAs and large hotel chains that open authenticated APIs will monetize provenance and push paid inclusion, creating a recurring data-licensing revenue stream and raising barriers for smaller proprietors.

Key risks that could reverse or compress these winners are non-linear: (1) a regulatory push for provenance/liability rules or “right to be omitted” for destinations (12–36 months) that raises compliance costs, (2) persistent hallucination incidents that shift consumers back to hybrid human-assisted models, and (3) political/regulatory backlash in over-touristed locales that curtails AI-driven demand allocation. Each catalyst has a timeline and would favor capital-light middleware and concierge providers over mass-market LLM-only players.

Tactically, the first wave of alpha will come from positioning around the data/API layer and compute stack while avoiding single-point exposure to hallucination liability. Trades should favor large platforms and cloud/GPU suppliers that can monetize structured access, while using options and pairs to defend against regulation or a UX-driven reversion to human agents.

Key Decisions for Investors

• Long Booking Holdings (BKNG) — 6–18 month horizon. Rationale: incumbent OTA with resources to monetize API/data access and sponsored inventory for LLMs. Positioning: buy stock or 9–12 month call spreads to cap premium; target asymmetric upside ~20–40% if AI-originated bookings rise while downside limited by strong core cash flow. Key risk: regulatory provenance rules or loss of consumer trust compressing take-rates.
• Long Nvidia (NVDA) — 3–12 month horizon via call spread on pullbacks. Rationale: GPU/inference demand from personalized travel planners should drive near-term utilization and pricing power. Risk/Reward: expect >2x upside on sustained enterprise LLM adoption; hedge with tight short call to fund position. Key risk: crypto-style demand cliff or broader market multiple compression.
• Pair trade: Long Expedia Group (EXPE) / Short Sabre (SABR) — 6–12 months. Rationale: EXPE can productize AI itineraries and capture booking volume; SABR (legacy GDS/distribution) is exposed if API-first routing bypasses GDS layers. Execution: equal notional sizes, trim if Sabre announces meaningful API modernization; asymmetry if distribution economics reprice in favor of platform-native flows.
• Long Microsoft (MSFT) cloud exposure — 9–18 months via LEAPS or 12-month call spreads. Rationale: enterprise ties to OpenAI and Azure position MSFT to capture higher-margin retrieval and fine-tuning services for travel verticals. Risk/Reward: moderate upside if travel verticals ramp; downside is broader multiple re-rating and competition from AWS/Google.