Glenn Fogel runs the world’s largest online travel business. He still personally reads customer complaints because ‘you gotta get that sh*t right’

Booking Holdings reported Q2 EPS of $2.54 (+15%) on revenue of $7.4B (+8%), beating Wall Street expectations, and the stock rose 6% on the release. Management attributed part of the improvement to AI reducing post-booking customer service friction, helping resolve issues faster. On the demand backdrop, international tourist numbers rose 2% in Q1 (Europe +4%), while strong tourism in America helped offset Middle East conflict headwinds.

Analysis

The real earnings signal is not that travel is holding up; it is that booking platforms are starting to convert service friction into margin. BKNG has a cleaner path than most consumer internet names to monetizing AI because every deflected customer-service contact is immediate opex relief, while the revenue side stays largely insulated by its transactional role. That makes this a quality/margin story more than a pure demand story, and it argues for multiple support if analysts start underwriting even modest EPS upside from lower support intensity.

Second-order winners are the companies with direct customer ownership and inventory scale; the losers are the intermediaries whose value is mostly discovery. Trip-planning AI can compress the economics of metasearch and lighter-weight travel apps faster than it harms the large OTA, because the incumbent already has supply connectivity, payments, and rebooking workflows. Over 6-18 months, the strategic risk is disintermediation by the interface owners: if booking starts inside a general AI assistant, BKNG must defend take rate with better fulfillment, not just brand.

Near term, the stock should trade on follow-through estimates rather than the macro travel print. The key falsifier is any sign that customer-service savings are offset by higher promotions, lower take rate, or softer international mix; if gross bookings keep growing but margins do not expand, the AI thesis is being oversold. The contrarian view is that consensus may already own the demand durability story, but still underappreciates the magnitude of opex leverage if AI reduces friction across cancellations, changes, and post-booking support.

Key Decisions for Investors

• Long BKNG on pullbacks over the next 2-4 weeks; base case is 5-10% upside as the market upgrades 2026 EPS/FCF from AI-driven service-cost reduction. Falsify on any guide that service expense per booking is flat or rising.
• Pair trade: long BKNG / short EXPE for 1-3 months. BKNG has better international exposure, stronger monetization of support automation, and less need to spend on top-of-funnel traffic. Cover if EXPE shows faster direct-booking growth or margin surprise.
• Do not chase the stock purely on travel demand strength; use next earnings to test whether AI is creating real opex leverage. If operating margin fails to expand despite healthy bookings, trim/exit because the market is paying for an efficiency narrative, not just volume.
• Watchlist only: short TRIP if AI assistants begin to reroute search traffic away from metasearch. This becomes actionable only if referral traffic or CPC trends deteriorate; until then it is an alert, not a position.