An independent case study

Booking Holdings: the OTA cash machine, meets the AI question

A neutral, evidence-first reading of Booking Holdings (NASDAQ: BKNG) — owner of Booking.com, Priceline, Agoda, Kayak and OpenTable — whose record 2025 results ran headlong into investor fear that Google and AI agents will disintermediate the online travel agency.

42 sourcesAs of 7 June 202611 analysis sections

In 2025 Booking Holdings did two things at once: it printed a record operating year — $186.1B gross bookings (+12%), $26.9B revenue (+13%), and more than 1.2 billion room nights[1] — and watched its stock fall ~23% on fears that AI agents would route travelers around it[31].

Booking is the largest online travel company on earth — roughly twice Airbnb and three times Expedia by night volume[18] — and a high-cash-generating business: $9.9B adjusted EBITDA at a 36.9% margin and $9.1B free cash flow in 2025, enough to fund a 25-for-1 split and ~$8.2B of buybacks and dividends[2][24]. Yet GAAP net income slipped 8% to $5.4B[2], and the market is pricing a structural question, not a cyclical one. The debate this case study lays out is genuinely two-sided — and the evidence is mixed on every side of it.

⚖️

The central question

Is Booking a durable travel toll-booth whose scale, supply and loyalty defend it from AI — or is it one layer of intermediation that a conversational agent can eventually collapse? The 2025 evidence points both ways: record fundamentals and AI partners that route through Booking, against a measurable shift of trip discovery into AI tools and a Google funnel that is getting more expensive.

Booking Holdings total revenue, 2019–2025 ($B) — pandemic crater to a 2025 record

The questions that decide the outcome

Four open questions organize this study. Each links to the section that weighs the evidence on both sides.

↳Will AI agents disintermediate the OTA — or route through it?

The fear that drove BKNG down ~23% in 2025: Google and ChatGPT/Gemini agents collapse the travel funnel into one chat box and bypass Booking. The counter: when OpenAI pulled back direct ChatGPT checkout, OTA stocks rose, and Google's named agentic hotel partners route through Booking.com, not around it. The contested middle ground is discovery — AI tools are increasingly where trips get planned, even if booking still lands on an OTA.

↳How durable is the Google-fed marketing machine?

Booking spent ~$7.3B (~31% of revenue) on marketing in 2024, with Google paid search the single largest line. Bull: app bookings exceeded 55% of room nights and Genius loyalty drives repeat direct demand, reducing reliance. Bear: Google's AI Overviews are cutting organic click-outs, forcing more paid spend just to stay visible.

↳Can the Connected Trip and Booking's own agents widen the moat?

Booking is building agentic products (Priceline's Penny, a 10+ agent system; Booking.com's AI Trip Planner) and pushing beyond hotels into flights and ground transport — Connected Trip transactions grew 20%+ in 2025. The open question is whether owning inventory, deals and trip context is a real AI advantage or a defensive cost.

↳How much does EU regulation erode the take rate?

Booking.com is the only travel company named an EU DMA gatekeeper, and Spain's CNMC fined it a record €413M for abusing a 70–90% local position. Price-parity clauses — long a pillar of the ~14.5% take rate — are being unwound. A real but gradual headwind, concentrated in Europe.

🤔

What reasonable people disagree about

Whether AI is a threat (discovery moves upstream into chat; the funnel collapses) or a tailwind (higher-conversion agents that still need Booking's supply); whether the ~14.5% take rate holds as the EU unwinds price parity; and whether ~8% room-night growth signals a maturing core or simply a larger base. This study presents both readings and leaves the judgment to you.

Overview & Timeline

From 'Name Your Own Price' to the world's travel toll-booth

How a dot-com bidding gimmick became, via an acquisition the trade press later called one of the most profitable in travel history, the largest online travel company on earth.

Founded 1996Norwalk, ConnecticutNASDAQ: BKNG

Booking Holdings is a portfolio of six travel brands — Booking.com, Priceline, Agoda, Kayak, OpenTable and Rentalcars — but its value is overwhelmingly the Booking.com hotel-and-homes engine acquired for ~$133M in 2005[13]. The company that bears Priceline's DNA is now run almost entirely on the European agency model it bought, not the bidding model it was born with.

What it is today

Booking Holdings is the parent of the world's largest online travel agency (OTA) business. Its flagship, Booking.com, listed roughly 4.4 million properties at the end of 2025 — about 500,000 hotels, motels and resorts and about 3.9 million homes and apartments[15]. Alongside it sit Priceline (US), Agoda (Asia-Pacific), Kayak (metasearch), OpenTable (restaurant reservations) and Rentalcars. The holding company is based in Norwalk, Connecticut, trades on the Nasdaq as BKNG, and has been led by CEO Glenn Fogel since 2017[10].

The deal that made the company

Priceline began as Jay Walker's 1996 "Name Your Own Price" venture and IPO'd in March 1999[10]. Its fortunes turned not on bidding but on Europe: it bought ActiveHotels for $161M in 2004 and Booking.com for roughly $133M in 2005, then merged them[11]. Booking.com's commission-based agency model and European hotel supply became the group's engine — a purchase later called one of the most profitable travel deals of the 2000s[13].

“Why Priceline's purchase of Booking.com was the most profitable travel deal of the 2000s.”

PhocusWire · travel-industry trade press · retrospective · source

Timeline

1996

Jay S. Walker founds Priceline in Connecticut around a 'Name Your Own Price' model. [10]

1999

Priceline.com IPOs in March; the William Shatner-fronted brand becomes a dot-com icon. [10]

2004

Buys ActiveHotels (Europe) for $161M — its first move into the agency hotel model. [11]

2005

Acquires Booking.com for ~$133M and merges it with ActiveHotels — the deal that remade the company. [13]

2007

Acquires Agoda, anchoring the Asia-Pacific business. [12]

2013

Buys Kayak (metasearch) for ~$1.8B. [12]

2014

Acquires OpenTable (restaurant reservations) for ~$2.6B; renames to The Priceline Group. [12]

2018

Renames to Booking Holdings, reflecting Booking.com's dominance of group value. [12]

2020

Pandemic: revenue collapses ~55% to $6.8B as global travel halts. [23]

2025

Record year: $186.1B gross bookings, $26.9B revenue, 1.2B+ room nights; 25-for-1 split announced. [1]

Strengths of the franchise

Six recognized brands across hotels, homes, flights, cars and restaurants, anchored by the dominant Booking.com supply base of ~4.4M properties [15].
An acquisition track record (Booking.com, Agoda, Kayak, OpenTable) that compounded into global scale [12].
Demonstrated resilience: a ~55% pandemic revenue collapse fully recovered to a 2025 record [23][1].

Limits of the franchise

Value is highly concentrated in one asset, Booking.com, and one region, Europe — a single point of regulatory and macro exposure [28].
The original Priceline bidding model is now a minor part of the story; growth depends on the agency/merchant hotel engine [14].
OpenTable and Kayak have never approached Booking.com's economics; the portfolio is top-heavy, not balanced.

Market & Value Chain

Where the money sits in online travel

The OTA is a two-sided marketplace between fragmented supply (hotels, hosts) and fragmented demand (travelers) — and it sits underneath an even more powerful layer: search and discovery.

Two-sided marketplace~14.5% take rate

OTAs earn by solving a fragmentation problem: millions of independent properties cannot each acquire global demand, and travelers will not search hundreds of hotel sites. Booking aggregates both sides and takes roughly 14.5% of the transaction[40]. The strategic risk is that the layer above Booking — Google, and now AI assistants — controls discovery and can re-price or re-route the demand Booking depends on.

The value chain

Travel distribution runs roughly: supply (hotels, hosts, airlines, car rental) → aggregation (OTAs like Booking; bed-bank wholesalers) → discovery (search engines, metasearch, AI assistants, social) → traveler. Booking occupies the aggregation layer and also owns a discovery asset (Kayak metasearch). The economics are best in aggregation — that is where the ~14.5% take rate lives[40] — but the poweris increasingly in discovery, because whoever owns the traveler's first click sets the price of demand.

A large, structurally growing market

Global travel is a multi-trillion-dollar industry and online penetration keeps rising, which is why Booking can post $186.1B of gross bookings and still grow room nights ~8% off a huge base[1]. Within accommodations, alternative accommodations (homes and apartments) are the fastest-growing segment, now about 36% of Booking.com room nights — the battleground with Airbnb[37]. Adjacent verticals — flights, ground transport, restaurant reservations — are smaller for Booking but are the target of its "Connected Trip" expansion.

🔗

The discovery chokepoint

Booking does not own demand outright. A large share of travelers begin on Google, and an increasing share begin in an AI assistant. Because Booking must buy much of that demand (its largest marketing line is Google paid search[17]), the discovery layer is both its biggest cost and its biggest strategic dependency — the thread that runs through this entire case study.

Why the market favors Booking

Supply and demand are both fragmented, the structural condition that makes a global aggregator valuable [15].
Online penetration of a trillion-dollar travel market keeps rising; the pie is still growing [1].
Booking's ~14.5% take rate on $186.1B of volume is an enormous, high-margin toll [40][1].

Why the market is precarious

The discovery layer (Google, AI assistants) sits above the OTA and can tax or re-route its demand [30].
Suppliers — especially large hotel chains — push direct booking to escape the take rate, and regulators now help them [34].
Take rates face downward pressure as price-parity clauses are unwound in Europe [27].

Business Model & Take Rate

How Booking turns bookings into cash

Two revenue models, one expensive habit. Booking has shifted from a commission collector to a payments-and-packaging merchant — lifting the take rate, but leaning ever harder on bought traffic.

Merchant ~66% of revenue~$7.3B marketing

Booking earns a blended ~14.5% take rate on gross bookings[40], increasingly via the merchant model (Booking handles payment, packaging and currency), which rose from ~24% of revenue in 2019 to about 66% in 2025[14]. The model is highly cash-generative — but its largest single cost is Google paid search, the crux of the disintermediation debate.

Agency vs. merchant

In the agency model, Booking connects a traveler to a hotel and collects a commission after the stay — capital-light, but limited. In the merchant model, Booking takes payment up front, bundles services (the "Connected Trip"), and earns on payments and packaging. The merchant share has flipped from roughly a quarter of revenue in 2019 to about two-thirds in 2025[14], lifting the effective take rate and giving Booking more control over the customer relationship.

Approximate FY2025 revenue mix by model (third-party estimate)
Merchant model (~66%) — 66%
Agency model (~30%) — 30%
Advertising & other (~4%) — 4%

Mix is a third-party estimate from a secondary analysis[14], not a Booking-disclosed segment split; treat as directional.

The marketing engine — and its dependency

Booking is one of the world's largest digital advertisers. In 2024 it spent about $7.3B on marketing — roughly 31% of revenue — with real-time bidding on Google search and metasearch the single largest category[17]. This is the heart of the bear case: a huge slice of customer acquisition flows through one channel (Google) that can raise prices, change its results page, or insert its own AI answers above Booking's paid links[30].

📲

The counterweight: direct and loyalty

Booking has deliberately reduced reliance on bought traffic. App bookings exceeded 55% of room nights in 2025, and the direct-channel mix held in the mid-50% range[16]. Its Genius loyalty program drives repeat, direct demand — levels 2 and 3 alone were ~50% of room share[26]. The more a traveler opens the app out of habit, the less Booking pays Google.

A high-margin cash machine

~14.5% take rate on $186.1B of volume converts into $9.9B adjusted EBITDA and $9.1B free cash flow [40][2][24].
The merchant shift (66% of revenue) deepens the customer relationship and adds payments economics [14].
Direct/app traffic (55%+ of room nights) and Genius loyalty cut dependence on paid search over time [16][26].

A model with a single chokepoint

~$7.3B (~31% of revenue) of marketing, concentrated in Google search, is a structural cost that Google partly controls [17].
Google's AI Overviews are reducing organic click-outs, forcing more paid spend to stay visible [30].
EU rules unwinding price parity let hotels undercut Booking on their own sites, pressuring the take rate [34].

Competitive Landscape

The biggest player in a contested oligopoly

Booking leads online travel globally, but it fights Expedia in the US, Airbnb in homes, Trip.com in Asia, hotels' own direct channels everywhere — and, increasingly, the discovery platforms above all of them.

Global #1 OTAFive Forces below

By night volume Booking is roughly 2x Airbnb and 3x Expedia[18], with structural strength in Europe and accommodations. But the most consequential competitive pressure is no longer another OTA — it is the discovery layer (Google, AI agents) and supplier-direct channels, both of which can intermediate or bypass the OTA.

Who Booking competes with

Expedia Group — #2 globally, strongest in the US, with a growing B2B (private-label) supply business; FY2025 revenue $14.73B[19][32].
Airbnb — dominant in alternative accommodations with a powerful direct brand and low paid-search reliance; FY2025 revenue $12.24B[20]. Booking counters by growing homes to ~36% of its room nights[37].
Trip.com Group — the leader in China and a fast international grower, ~$48B market cap[22].
Supplier-direct & Google — large hotel chains push direct booking, and Google/AI assistants control discovery and can route demand[30].

Five Forces

The defining feature is that the two highest-pressure forces — rivalry and substitutes — both point at the same thing: the cost of acquiring demand. Click a force to see the basis.

Online travel agencies

Competitive rivalry — High. Online travel is a concentrated but fiercely contested oligopoly. Booking leads globally (especially Europe) with $186.1B gross bookings, but competes head-on with Expedia (#2, US-strong), Airbnb (dominant in alternative accommodations) and Trip.com (China/Asia). Competition is fought largely on marketing spend and supply breadth, which keeps the customer-acquisition arms race expensive.

Positioning: supply scale vs. owning demand

The map below plots competitors on two axes that matter in an AI world: supply breadth/scale (vertical) and how much each player owns its own demandvs. depends on others' traffic (horizontal). Booking is the largest on scale but only mid-pack on owning demand; Airbnb is smaller but owns its demand; Google/AI agents own demand outright with almost no supply — which is exactly why the threat comes from the demand side.

Online travel: supply scale vs. control of demand (qualitative)

Hover a point to see the basis for its placement.

Booking's competitive edge

Unmatched scale: ~2x Airbnb and ~3x Expedia by night volume, with the deepest accommodations supply [18][15].
Genius loyalty and 55%+ app bookings build a repeat, direct demand base that rivals struggle to match [26][16].
Closing Airbnb's gap: alternative accommodations now ~36% of Booking.com room nights [37].

Where rivals press

Airbnb owns its demand with a stronger consumer brand and far lower paid-search reliance [20].
Trip.com dominates China, a market where Booking is structurally weak [22].
The real squeeze is the discovery layer: Google and AI agents sit above every OTA and set the price of demand [30][7].

The Central Debate

OTA cash machine vs. AI-agent disintermediation

The single question driving Booking's valuation: will Google and conversational AI agents route travelers around the OTA — or will they keep routing through it? The 2025 evidence cuts both ways.

BKNG −23% in 2025Discovery vs. booking

In 2025 the stock fell ~23% on fears that AI agents would bypass OTAs[31]. Two facts complicate the panic: when OpenAI scaled back direct ChatGPT checkout, OTA stocks rose[5], and Google's named agentic hotel partners route through Booking.com, not around it[6]. But the share of travelers using AI tools "extensively" for trip planning more than doubled (13% → 30%)[7] — so the threat may be migrating from booking to discovery.

The bear case: the funnel collapses into a chat box

The disintermediation thesis is concrete. If a traveler can say "book me four nights in Lisbon under €200 with a pool" to an AI agent and have it transact directly with hotels, the OTA's aggregation role — and its ~14.5% take rate — is squeezed out. In February 2026, Citizens downgraded BKNG to Market Perform, warning explicitly that AI could collapse the funnel and push Booking down the value chain.

“AI could eventually transform the online travel agency (OTA) model by collapsing the traditional travel funnel into a single conversational interface ... push companies like Booking Holdings further down the value chain.”

Citizens · equity research (downgrade to Market Perform) · 4 Feb 2026 · source

Compounding it: Google's AI Overviewsare reducing organic "click-outs" to Booking's site, which has pushed Booking to spend more on sponsored links just to stay visible[30]. Even if booking stays with the OTA, the discoveryshift — 13% to 30% of travelers planning "extensively" in AI tools[7] — raises the cost of demand.

The bull case: agents need supply, deals and context

The counter-argument is that an AI agent is only as good as the inventory, pricing and trust it can connect to — and Booking owns that. Two market signals support it. First, OpenAI retreated from native ChatGPT checkout (only a handful of merchants went live), and OTA stocks rallied because early AI travel products mostly direct traffic to OTA apps and sites rather than transacting themselves[5]. Second, when Google built agentic hotel booking, its partner list was the incumbents.

“Agentic commerce was supposed to route around the OTAs. Google's partner list routes straight through them.”

Hospitality Net · industry analysis on Google's agentic-booking partners · 2026 · source

Some analysts go further, arguing disintermediation risk is "highly likely to be overstated for OTAs, especially for Booking", given its supply relationships and scale[38]. And Booking is hedging every way at once — partnering with Google, OpenAI, Anthropic and Amazon simultaneously while building its own agents (Penny, a Booking.com agent, and a stealth venture, Lola)[29].

🧭

The synthesis: discovery is the real battleground

The near-term evidence (2025–26) favors the bull case on transactions: agents are routing through OTAs, not around them[6]. The bear case is strongest on discovery: more trip-planning is starting in AI tools, which can raise Booking's cost of demand even if it keeps the sale[7]. The unresolved question is whether, over several years, owning supply and loyalty is enough to keep Booking the place the agent transacts — or whether the agent becomes the new toll-booth above it.

Share of travelers using AI tools 'extensively' for trip planning (YoY)

The discovery battleground, quantified: the share of travelers planning "extensively" in AI tools more than doubled year-over-year, from 13% to 30% (the source reports the pair as year-over-year without naming the survey dates)[7]. Transactions still largely route through OTAs today[6], so this discovery shift — not checkout — is the live edge of the disintermediation debate.

Why AI may not disintermediate Booking

OpenAI scaled back direct checkout; OTA stocks rose as AI products direct traffic to OTAs rather than transacting [5].
Google's agentic hotel-booking partners route through Booking.com and other incumbents, not around them [6].
Agents need supply, real-time pricing, deals and trust — Booking's assets; some analysts call the risk overstated [38].
Booking is partnering with every major AI platform while building its own agents, hedging the outcome [29].

Why AI may erode Booking

Trip-planning is shifting into AI tools fast — "extensive" use rose from 13% to 30% YoY [7].
Google's AI Overviews cut organic click-outs, forcing Booking to spend more on paid links [30].
Citizens warns AI could collapse the funnel into one chat interface and push OTAs down the value chain [4].
If the agent owns the user relationship, it — not Booking — sets the price of demand long-term.

Strategy: Connected Trip & AI

Booking's answer: own the agent, own the trip

Booking's stated strategy is to turn the AI threat into a moat — by building its own agentic assistants on top of proprietary inventory and deals, and by expanding from hotels into a full 'Connected Trip'.

Penny: 10+ agentsConnected Trip +20%

Booking's thesis is that the AI advantage comes not from the model but from connecting it to inventory, deals and trip context — which Booking owns[9]. It is executing on two fronts: its own agentic assistants(Priceline's Penny, a 10+ agent system; Booking.com's AI Trip Planner) and the Connected Trip, where multi-component transactions grew 20%+ in 2025[25].

Stated strategy vs. revealed strategy

What management says:CEO Glenn Fogel calls advancing generative AI a key priority "to enhance the value we deliver to both travelers and partners"[3]. The 10-K describes AI trip planners, AI assistants and a price-comparison tool across brands, plus work with leading AI firms on "agentic developments"[35].

What it actually shipped: on 3 June 2026, Priceline launched a fully agentic next-generation Penny— described as "a coordinated system of more than 10 specialized agents" that takes a user "from trip idea to booking in one conversation" across hotels, flights, cars and service[8]. The revealed strategy matches the stated one: build the agent in-house rather than cede the interface to a third party.

“The real advantage in AI travel would not come from the model alone, but from connecting it to the context, inventory and deals.”

Sejal Amin · CTO, Priceline · 3 Jun 2026 · source

🤝

An Anthropic partner makes the same argument

Anthropic's Guillaume Princen, a partner on Penny, framed it identically: "The value of AI in travel comes from connecting models to the real context of inventory, deals, and trip purpose."[42] Whether that proprietary context is a durableadvantage — or something agents can eventually replicate via APIs and partnerships — is the strategy's central bet.

The Connected Trip and the Genius flywheel

Beyond AI, Booking is widening the relationship. Connected Trip transactions — combining accommodations with flights, ground transport and more — grew over 20% in 2025, with flights a major contributor[25]. The aim is retention: the more verticals a traveler books in one place, the less likely an agent can peel them away. Reinforcing it, the Geniusloyalty program's top tiers (levels 2–3) were over 30% of the active base and ~50% of room share in 2025; those members book more often, earlier, and return more consistently[26].

Why the strategy can work

Booking owns the supply, real-time pricing and deals an agent needs — the asserted durable advantage [9][42].
It is shipping real agentic product (Penny, 10+ agents) rather than just talking about AI [8].
Connected Trip (+20%) and Genius loyalty (~50% room share) deepen retention against disintermediation [25][26].

Why it may not be enough

Building agents is also a defensive cost; success is unproven and the interface war is early [4].
If consumers default to a general assistant (ChatGPT/Gemini), Booking's own agent may struggle for attention [7].
Proprietary inventory can be exposed to third-party agents via APIs — context may be less defensible than claimed.

Regulation & Antitrust

The price of dominance in Europe

Booking's strength in Europe is also its biggest regulatory exposure. It is the only travel company designated an EU 'gatekeeper', and Spain handed it a record fine for abusing a dominant position.

€413M Spain fineDMA gatekeeper

Two 2024 actions reshape Booking's European economics: the EU named Booking.com a DMA gatekeeper (the only travel company on the list)[39], and Spain's CNMC fined it a record €413M for abusing a 70–90% local position[28]. Both target the same lever — price-parity clauses— that long protected Booking's take rate. The effect is real but gradual, and concentrated in Europe.

The DMA gatekeeper designation

On 13 May 2024 the European Commission designated Booking.com a gatekeeper under the Digital Markets Act — the seventh, joining Alphabet, Amazon, Apple, ByteDance, Meta and Microsoft[39]. From 14 November 2024Booking must comply with DMA obligations, including a ban on wide price-parity clauses. In practice, hotels and other providers are now "free to offer different (including better) prices and conditions on their own website or other channels" than on Booking.com[34].

⚠️

Why parity matters

Price-parity clauses prevented hotels from undercutting Booking on their own sites — which protected Booking's take rate and discouraged direct booking. Unwinding them lets suppliers compete on price against the OTA, a structural (if slow-moving) pressure on the ~14.5%take rate in Europe, Booking's strongest region[40].

Spain's record fine

In July 2024 Spain's competition authority (CNMC) fined Booking.com €413.24M — the largest fine in its history — for abusing a dominant position between 2019 and 2024[28]. The CNMC found Booking held a 70–90% share of online booking intermediation for Spanish hotels and used unfair terms plus a search-ranking system that pushed hotels to concentrate bookings on its platform[28]. Booking can appeal, and the financial hit is modest against $9.1B of free cash flow — but the precedent and the behavioral remedies matter more than the cash.

Why regulation is manageable

The €413M fine is small against $9.1B of annual free cash flow; it is absorbable [24][28].
Direct/app traffic (55%+ of room nights) and loyalty reduce dependence on the parity-protected channel [16].
Regulation hits all large OTAs operating in Europe, not Booking uniquely on the demand side.

Why regulation is a real drag

Unwinding price parity lets hotels undercut Booking directly, pressuring the take rate in its strongest region [34].
DMA gatekeeper status brings ongoing compliance obligations and scrutiny that lighter rivals avoid [39].
A 70–90% national share invites more antitrust action across the EU, not just Spain [28].

Financials & Growth

A record year — with a profit asterisk

2025 set records for gross bookings, revenue and cash flow, and funded a 25-for-1 split. Yet GAAP net income fell, and room-night growth has settled into the high single digits.

$26.9B revenue$9.1B FCFNet income −8%

2025 was the best operating year in Booking's history on the metrics that compound — $186.1B gross bookings (+12%), $26.9B revenue (+13%), $9.9B adjusted EBITDA at a 36.9% margin, and $9.1B free cash flow[1][2]. The asterisks: GAAP net income fell 8% to $5.4B[2], and room-night growth has moderated to ~8%[1].

Gross bookings: the compounding base

Gross bookings — total transacted travel value — is the cleanest gauge of platform scale. It has roughly doubled since 2022, reaching $186.1B in 2025[1].

Booking Holdings gross travel bookings, 2021–2025 ($B)

The 2025 profit picture

The headline tension: revenue, adjusted EBITDA and free cash flow all hit records, but GAAP net income fell 8% to $5.4B[2]. The gap reflects comparison effects and cost items below the EBITDA line; on the cash-generation that funds buybacks, 2025 was a step up. Q4 alone delivered $6.3B revenue (+16%), $43.0B gross bookings (+16%), 285M room nights (+9%) and $1.4B GAAP net income (+34%)[36].

FY2025 headline financials ($B) — records on cash, a dip on GAAP net income

Revenue

$B26.9

Adj. EBITDA

$B9.9

Free cash flow

$B9.1

GAAP net income

$B5.4

💰

Capital returns and the split

2025 free cash flow of $9.1B (+15%) and adjusted EPS of $228.06 (+22%) underwrote heavy shareholder returns: the board raised the dividend 9.4% to $10.50/share and approved a 25-for-1 stock split effective 2 April 2026[24]. Including buybacks, Booking returned billions to shareholders — the capital-return profile of a mature, cash-generative company.

The financial bull case

Records across gross bookings, revenue, adjusted EBITDA (36.9% margin) and free cash flow ($9.1B) [1][2][24].
Prodigious cash returns: rising dividend, large buybacks and a 25-for-1 split [24].
Q4 acceleration (revenue +16%, net income +34%) suggests momentum, not fade [36].

The financial bear case

GAAP net income fell 8% even in a record year, a reminder the bottom line isn't only rising [2].
Room-night growth has decelerated to ~8% — a larger base, but slower compounding [1].
Heavy buybacks flatter per-share metrics; the de-rating reflects a structural, not cash-flow, worry [31].

Peer Comparison

Booking vs. Expedia, Airbnb and Trip.com

Booking is the scale leader and prints peer-leading margins, yet trades at a lower forward multiple than Airbnb — the market paying up for Airbnb's owned demand, not Booking's larger book.

Scale leader~15.6x fwd P/E

By revenue, gross bookings and margin, Booking leads its OTA peers — FY2025 revenue of $26.9B versus Expedia's $14.7B and Airbnb's $12.2B[19][20]. Yet it trades at roughly 15.6x forward earnings against Airbnb's ~25.4x[21] — the valuation gap is the AI-disintermediation and Google-dependency discount in one number.

Scale: revenue by player

FY2025 revenue, major OTAs ($B) — Booking ~2x Expedia, ~2.2x Airbnb

Booking Holdings

$B26.9

Expedia Group

$B14.73

Airbnb

$B12.24

Trip.com (est.)

$B8.5

The comparables table

Company (FY2025)	Revenue	Gross bookings / GBV	Room nights / nights	Net income	Fwd P/E*
Booking Holdings	$26.9B	$186.1B	1,235M	$5.4B	~15.6x
Expedia Group	$14.73B	$119.59B	415.4M	$1.29B	~11.8x
Airbnb	$12.24B	$91.3B (GBV)	533.0M	~$2.6B (est.)	~25.4x
Trip.com Group	~$8.5B (est.)	—	—	—	—

Booking, Expedia and Airbnb FY2025 figures from filings / filing-summary services[1][19][20]. Trip.com revenue is an estimate[22]. *Forward P/E as of early April 2026[21]. Airbnb net income is an approximate derivation from its disclosed ~21% net margin[41].

🏷️

Why does the leader trade at a discount?

Booking posts a slightly higher 2025 adjusted EBITDA margin than Airbnb (36.9% vs ~35%) on a far larger booking base, yet trades at a much lower forward multiple[41][21]. The market is paying for Airbnb's owned demand(strong direct brand, low paid-search reliance) and discounting Booking's Google dependency and AI risk — the same threads from the central debate, expressed as a P/E gap.

Booking screens cheap

Largest scale and peer-leading margins, but a lower multiple than Airbnb — a value setup if AI fears fade [41][21].
Far bigger gross bookings ($186.1B) and free cash flow fund buybacks that compound per-share value [1][24].
Closing Airbnb's homes edge (~36% of room nights) narrows the differentiation gap [37].

The discount may be deserved

Airbnb's owned-demand model is structurally less exposed to Google/AI intermediation — worth a premium [20].
Expedia trades even cheaper (~11.8x), so "cheap vs Airbnb" isn't unique to Booking [21].
If AI disintermediation is real, the leader with the most paid-traffic exposure has the most to lose [30].

Risks (SWOT)

What could break the thesis

An even-handed SWOT: Booking's scale and cash generation are not in question. The contested items are all on the right-hand side — disintermediation, Google dependency, regulation and decelerating growth.

Threats weighted equally

The strengths are concrete and current (scale, margin, loyalty); the threats are structural and forward-looking (AI/agents, Google, EU regulation). A reasonable reader can accept every strength below and still worry about every threat — which is precisely why the stock de-rated in a record year.

Strengths

Scale leadership: $186.1B gross bookings, $26.9B revenue (+13%) and 1.2B+ room nights in 2025 - roughly 2x Airbnb and 3x Expedia by night volume (s1, s18).
Cash machine: $9.9B adjusted EBITDA (36.9% margin) and $9.1B free cash flow funded ~$8.2B of buybacks/dividends and a 25-for-1 split (s2, s24).
Direct/loyalty moat: app bookings >55% of room nights and Genius levels 2-3 driving ~50% of room share - a repeat-customer flywheel that reduces paid-traffic reliance (s16, s26).

Weaknesses

Google dependency: ~$7.3B (~31% of revenue) marketing spend, with Google SEM the largest category - a single powerful channel that can re-price its own auction (s17).
GAAP net income fell 8% in 2025 even at record sales, and room-night growth has decelerated to ~8% (s2, s1).
Concentration in Europe and accommodations leaves it more exposed to EU regulation and regional macro shocks (s27, s28).

Opportunities

Connected Trip: non-accommodation verticals (flights, ground transport) grew 20%+ in 2025 and deepen retention (s25).
Owning the AI layer: agentic Penny (10+ agents) and Booking.com's AI Trip Planner can turn AI from threat into a higher-conversion funnel (s8, s35).
Alternative accommodations now ~36% of Booking.com room nights, narrowing Airbnb's historical edge (s37).

Threats

AI/agent disintermediation: travelers increasingly plan in ChatGPT/Gemini before reaching an OTA (13%->30% YoY); Citizens warned the funnel could collapse into one chat interface (s7, s4).
Regulation: DMA gatekeeper obligations and Spain's record EUR413M fine unwind price-parity and pressure the take rate (s27, s28, s34).
Google's AI Overviews are cutting organic click-outs, forcing more paid spend to keep visibility (s30).

🚩

The risk that matters most

Every other risk is manageable against $9.1B of free cash flow. The one that isn't fully in Booking's control is discovery moving into AI: if travelers increasingly start — and eventually transact — inside a general-purpose agent, Booking's cost of demand rises and its toll-booth position weakens[7][4]. Booking's own agents and Connected Trip are the bet that it can stay the place the transaction lands[8].

Methodology & Limits

How this was built — and where it may be wrong

A neutral, source-first compilation assembled from primary filings, regulator statements, company press releases and reputable trade press, with an explicit accounting of estimates and uncertainty.

42 sourcesAs of 7 June 2026

Approach

This case study was produced by fan-out web research and source-by-source verification. Every cited URL was fetched or read during the research run; figures were preferred from primary sources (Booking, Expedia and Airbnb results; EU Commission; PR Newswire) and corroborated with reputable trade press (PhocusWire, TechCrunch) and filing-summary services. The goal is a balanced compilation that lets you weigh the OTA-cash-machine-vs-AI-disintermediation question yourself — not an argument for either side.

Neutrality & balance

Each section presents the case for and against and attributes critical claims to named sources. The source base is roughly balanced across stance — supporting, critical and neutral — and the Sources page shows the exact mix. Where evidence leans one way (e.g. that AI agents currently route through OTAs), the strongest counter is shown alongside.

🧭

Where this case study may be wrong

Estimated figures. The merchant/agency revenue split (~66%/30%) and the ~14.5% take rate are third-party estimates, not Booking-disclosed segment data[14][40]. Trip.com revenue and Airbnb's implied net income are estimates[22][41].
Bot-walled primaries.SEC.gov filing pages and Skift blocked automated fetching; headline financials here come from filing-summary services (StockTitan), the company's own press release figures and trade press that were fetched and read — but were not cross-read against the raw 10-K line items.
A fast-moving debate.The AI-disintermediation picture (OpenAI's checkout walkback, Google's partner list, analyst stance) is evolving month to month; this is a 7 June 2026 snapshot[5][6].
The 13%→30% AI-planning stat comes from a single survey cited in trade press; treat as directional, not definitive[7].

Independence

This is an independent case study. It is not affiliated with, endorsed by, or sponsored by Booking Holdings or any company mentioned. All trademarks belong to their owners. Nothing here is investment advice — no rating, price target, or recommendation to buy or sell any security. Figures are as of 7 June 2026 and will go stale; verify before relying on any number.

Bibliography

Sources

Every cited source was fetched or read during the research run. Tiers: 1 = primary/official (Booking & peer SEC filings, EU Commission, company press releases), 2 = reputable press/research (TechCrunch, PhocusWire, Investing.com, StockTitan filing summaries), 3 = tertiary (market-data sites, analyst Substacks, encyclopedic write-ups).

42 sources

Tier 1: 7Tier 2: 16Tier 3: 19·Supporting: 19Critical: 8Neutral: 15

Executive Summary

[1]StockTitan — Booking Holdings Q4 2025 results & 25-for-1 split (8-K summary) T2 supporting en
Booking Holdings reported 2025 revenue of $26.9B (+13%), gross bookings of $186.1B (+12%), and over 1.2B room nights (+8%).
[2]StockTitan — Booking Holdings Q4 2025 results (8-K summary) T2 neutral en
2025 GAAP net income was $5.4B, down 8% year over year, while adjusted EBITDA rose 20% to $9.9B (36.9% margin).
[3]StockTitan — Booking Holdings Q4 2025 results (8-K, Fogel quote) T2 supporting en
CEO Glenn Fogel framed advancing the company's use of generative AI as a key priority.
[4]MarketMinute — Booking Holdings Q4 2025 earnings: navigating the 'AI scare trade' T3 critical en
BKNG fell ~23% year-to-date in 2025 amid an 'AI scare trade' on fears that Google/OpenAI agents might bypass OTAs.
[5]PhocusWire — Booking Holdings Q4 and full-year 2025 earnings T2 supporting en
2025 marked a record operating year for Booking even as the stock de-rated on AI fears, setting up the central OTA-cash-machine-vs-disintermediation debate.