Travel Through the Lens of AI with Booking.com CEO Glenn Fogel

2026-07-09 - 41 min - source - Read full transcript

Elad Gil (host)Glenn Fogel

Key insights

Fogel argues no company has a durable moat; the only long-term strategy is continuous reinvention.

He repeats this point twice in the conversation: today's competitive advantages in areas like alternative accommodations can disappear tomorrow, so Booking's approach is to keep developing new services and better ways of serving travelers and partners rather than assuming scale alone protects the business.

ai-and-competitive-moats

Scale and regulatory complexity are underappreciated moats that new AI-native entrants underestimate.

Fogel says building a travel marketplace isn't just aggregating inventory into a database - it requires thousands of people managing relationships with hotels and property managers, plus navigating an increasingly complex global regulatory framework for merchants of record in travel, which large incumbents can absorb but new entrants often don't understand going in.

ai-and-competitive-moats

OpenAI's brief launch and cancellation of a ChatGPT checkout/travel feature reassured travel incumbents that AI labs are not planning to disintermediate them as merchants.

Fogel says the market read OpenAI's decision not to become a merchant of record or maintain an in-app commerce flow as a sign the labs are stepping back from direct competition with existing travel platforms; Booking's stock rose about 8% on the news of the feature's cancellation.

agentic-commerce-in-travel

Fogel sees AI agents as most valuable for complex, multi-constraint trips rather than simple bookings.

He describes using Penny to plan a multi-city family trip with different cabin classes, mixed frequent-flyer-mile and cash payment across travelers, and sequencing decisions about hotels versus onward travel - the kind of dense, iterative planning that used to require a human travel concierge and that most travelers currently just avoid doing well.

agentic-commerce-in-travel

Most travelers still want some agency in the decision, not full automation, especially for complicated trips.

Fogel distinguishes simple errands (like a one-way business flight) where full automation on a trusted assistant's behalf is fine, from complex family trips where people want to review and confirm options themselves; Booking's agent design keeps a human confirmation step for complex itineraries.

agentic-commerce-in-travel

Booking is running AI/agentic usage against a hard ROI bar, tracking token cost per interaction against downstream value.

Fogel says Penny's adoption has been doubling roughly monthly, but the company is deliberately not pushing it harder yet because it wants to understand the true cost (tokens consumed, back-and-forth turns, which model for which task) versus ROI signals like conversion, faster search-to-booking, lower cancellation rates, and long-term customer lifetime value before scaling further.

agentic-commerce-in-travel

AI-driven customer service has already cut Booking's cost per contact by about 10% while improving satisfaction, but full automation isn't the goal.

Fogel cites Booking's Q1 earnings call figure of roughly 10% lower cost per reservation-related contact from AI, and says satisfaction is up too, but stresses that some customers still want a human, so the design goal is matching the channel to what's best for the customer rather than maximizing AI usage for its own sake.

agentic-commerce-in-travel

Capital allocation should default to returning cash to shareholders unless a reinvestment or acquisition clears a positive-ROI bar.

Fogel frames Booking's roughly $700 million 2025 tech/AI-adjacent investment and its record buyback and dividend activity (about $3.6 billion in buybacks and $300+ million in dividends in one quarter cited, plus roughly 40% of outstanding shares repurchased over about twelve years) as following a simple rule: only keep cash in the business if it can generate sufficient ROI internally or via acquisition, otherwise return it because shareholders can deploy it better themselves.

capital-allocation-and-scale

The speed of AI-driven job displacement, not displacement itself, is the real societal risk.

Fogel points to Booking's own history - hotel content and customer service used to require humans working across 40 languages before machine translation eliminated most of that work - as proof that technology has always displaced jobs, but argues the current AI wave is unusually fast, and that new-job creation may not keep pace with job loss for people like an established truck driver who can't easily retrain.

ai-and-the-future-of-work

Company-led upskilling is Fogel's answer to AI job displacement, though he is skeptical government retraining programs will help.

He describes ongoing internal conversations with his HR/people leadership about training employees to use AI tools so they remain competitive even if their current role changes, framing it as both good for the company (more productive workers) and an obligation to employees, while noting that government retraining efforts over the last 50 years have a poor track record.

ai-and-the-future-of-work

Public opinion on AI flips dramatically depending on how the question is framed, which distorts the real policy conversation.

Fogel and Gil discuss survey research showing that asking people if they enjoy using tools like ChatGPT, Gemini, or Claude produces strongly positive responses, while asking if they fear AI will destroy their career produces strongly negative ones; both argue this framing effect, amplified by motivated actors in a democracy, gets in the way of an honest conversation about real costs and benefits across the economy.

ai-and-the-future-of-work

Fogel's core life advice is to choose a career path deliberately, not by inertia, because time can't be recovered.

Reflecting on law-school classmates who defaulted into legal careers because it was the expected path and later felt wistful about roads not taken, Fogel argues that anyone with some freedom of choice should choose their path wisely, since a person only gets one life and can't get the time back.

career-and-life-choices

Companies

Booking Holdings - Fogel's company (formerly Priceline); grew from a few hundred million dollars in market cap in 2000 to over $100 billion, closing 2025 with 8.6 million alternative-accommodation listings and roughly $186 billion in gross travel bookings.
Priceline - The original name of the company Fogel joined in 2000; its AI trip-planning assistant is called Penny.
OpenAI - Discussed for briefly launching and then canceling a checkout/travel-booking feature in ChatGPT, and for the debate over whether AI labs will disintermediate merchants like Booking.
Anthropic - Cited alongside OpenAI as rumored to have a $30-50 billion annual revenue run rate, used as a comparison point for the scale of the current AI boom versus the dot-com boom.
Airbnb - Compared to Booking Holdings' alternative-accommodation business; Fogel says globally the two are close in listings and Booking is about three-quarters of Airbnb's size in that segment alone, and has grown faster than Airbnb there over the last five years.
Decagon - Cited by Elad Gil as an example of a company building agents specifically for customer support.
Google - Cited alongside OpenAI and Anthropic as a large platform pushing increasingly self-driven, agentic AI systems.

Techniques and frameworks

Penny (Priceline's AI agent) - Priceline's agentic trip-planning system; Fogel describes using it personally to plan a complex multi-city, multi-generation family trip, including optimizing which travelers use frequent-flyer miles versus cash.
No-moat philosophy - Fogel's recurring framing that no competitive advantage is permanent; the only durable strategy is continuously building new services and improving execution for travelers and partners rather than assuming past scale protects the business.
Token-cost ROI accounting - Fogel describes tracking the cost of running AI agents like Penny in terms of tokens consumed per interaction, weighed against downstream ROI (conversion, cancellation rate, lifetime value) before scaling usage.

Summary

Elad Gil interviews Booking Holdings CEO Glenn Fogel, who joined the company (then Priceline) in 2000 when it was worth a few hundred million dollars, one week before the Nasdaq peaked and sent the stock briefly toward delisting. Fogel walks through his path from a Wall Street trading job he disliked, through a personal low point (a job loss, his father's death, and a failed attempt to get a novel published) that led him into Priceline's corporate development team, and 25-plus years later into running a company that scaled roughly 1,000x to over $100 billion in market value. He draws a direct parallel between the dot-com boom and today's AI boom: both produced a handful of companies with enormous real revenue (he cites rumored $30-50 billion run rates at OpenAI and Anthropic) alongside many highly valued but pre-revenue companies, and he expects a similar pattern of most entrants failing while a few durable winners emerge, as has happened repeatedly across speculative booms from the California gold rush onward.

Much of the conversation centers on how AI is reshaping travel specifically. Fogel pushes back on the idea that large language models will let AI labs or new entrants simply disintermediate travel companies, pointing to OpenAI's decision not to become a merchant of record and its cancellation of a ChatGPT checkout/travel feature (which sent Booking's stock up about 8%) as evidence the outside perception of easy disruption was overstated. He argues durable advantages in travel come from operational depth most outsiders don't see: thousands of staff managing partner relationships, and an increasingly complex global regulatory framework for travel merchants that big, well-capitalized incumbents can absorb but new entrants often underestimate. At the same time, he is enthusiastic about AI as a tool, describing in detail how he personally used Priceline's agentic assistant Penny to plan a complex multi-city, multi-generation family trip, mixing frequent-flyer miles and cash payment across travelers. He frames the company's approach as keeping humans "in the loop" via confirmation for complex trips while allowing more automation for simple ones, and says the company is deliberately measuring the token cost of every agent interaction against ROI signals like conversion, cancellation rate, and lifetime value before scaling adoption further, even though Penny usage has been roughly doubling month over month.

On capital allocation, Fogel describes a simple internal rule: only reinvest or acquire if the expected ROI clears the bar, otherwise return cash to shareholders, who cited roughly $700 million in 2025 tech/AI-adjacent investment against a record quarter of buybacks and dividends, and roughly 40% of outstanding shares repurchased over about twelve years. He also makes the case that Booking's alternative-accommodation business is a much bigger, more durable asset than commonly assumed, roughly three-quarters the size of Airbnb globally and growing faster over the past five years, on top of Booking's larger hotel business.

The conversation closes on the societal side of AI: Fogel argues job displacement from technology is not new, citing Booking's own history of eliminating large human customer-service and translation teams once machine translation matured, but says the unusual speed of the current wave, and the risk that new job creation won't keep pace with the losses for workers like an established truck driver, deserves more serious public conversation than it's getting. He describes internal upskilling efforts as the company's own answer, is skeptical that government retraining programs have a good track record, and flags that public survey data on AI sentiment flips dramatically based on how questions are framed, which he thinks distorts the broader policy debate. He ends with reflections on choosing a life path deliberately rather than by default, informed by watching law-school classmates settle into careers by inertia and later feel wistful about it.

Notable Quotes

"There is no such thing as a moat. There is no such thing... The only way to win long term is to continue to develop new services, new ways to do things." - Glenn Fogel

"If you think you're just going to come in and do this business and knock away these very big players, I'd say you should really understand what the business is before you decide to commit your capital." - Glenn Fogel

"Choose wisely, because you will not get that time back." - Glenn Fogel

"The truth is, the way we look at AI is an incredible, beneficial tool, and a way for us to be able to do our mission easier, cheaper, and better for our customers." - Glenn Fogel