A near-perfect business with one very big problem

Today, I'm digging into Booking Holdings (BKNG).

They run a variety of online travel marketplaces: Booking.com for hotels, Priceline, Agoda in Asia, KAYAK for price comparison, Rentalcars.com, and OpenTable for restaurant reservations.

Similar to Autozone, Booking runs on negative working capital.

When you book a stay on many of their properties, they collect your money at the time of booking, but they often don't pay the hotel until after you've checked out. That timing gap means Booking is holding a giant pile of other people's cash at all times. Strip out their own cash and debt, and the money they owe (payments due to hotels, deferred bookings, accrued expenses) exceeds their operating current assets by more than $10B.

In today’s interest rate environment, that float is a heck of a feature of this business model. It funds the business, and it grows automatically as bookings grow. It's the same dynamic that makes insurance floats and Costco memberships so powerful: get paid first, pay later.

A few things that stood out to me:

• Negative working capital turns growth into a self-funding machine. Every incremental booking hands Booking cash before it owes anyone. The faster it grows, the more free financing it generates, which is the opposite of most businesses that have to fund growth with working capital.

• Operating margins went from 27% in FY2023 to roughly 34% in FY2025 while revenue climbed from $21.4B to $26.9B. That's operating leverage on an asset-light model that owns almost no hard assets. Free cash flow was about $9B last year on $322M of capex.

But it's not all clean and perfect:

• Their distribution is largely rented. A big chunk of traffic comes through Google, which is both the landlord Booking pays rent to and a competitor building its own travel products. Now AI agents increasingly sit between Booking and the traveler too.

• The stock is down about 24% over the past year even as the business grew nicely, mostly on fear that AI chat assistants will disintermediate the whole online travel model.

I wanted to highlight another business with negative working capital after my AutoZone report, and I was planning on that being the key takeaway, but I was way off.

AI is far from perfect. I find myself using tools like ChatGPT and Claude almost exclusively these days when it comes to analyzing purchase decisions. Similar to how Google moved directly into the travel space with its Google Flights feature, I would be shocked if large AI companies don't try to roll their own travel offerings.

Booking is one of the most impressive businesses I've come across, but it's also in one of the most precarious positions I've seen. I’ve now put their ticker on my watchlist as this is a space I’m excited to see play out!

With that, I'll see you in a couple of days.

Nick

TL;DR

• Booking Holdings runs the world's largest set of online travel marketplaces (Booking.com, Priceline, Agoda, KAYAK, Rentalcars, OpenTable) and earns a commission on reservations.

• It makes money as an asset-light middleman, and its structurally negative working capital means it collects from travelers before paying hotels, funding growth with float.

• Recent performance is strong: revenue up to $26.9B in FY2025 with operating margin near 34% and roughly $9B of free cash flow.

• The financial profile is rare: high margins, tiny capex, big buybacks (book equity is negative because of it), and modest net debt of about $2.1B.

• Stock story: shares are down about 24% over the trailing year despite growth, lagging Expedia (up about 35%) while roughly tracking Airbnb, as the market handicaps AI disruption risk.

• Operator takeaway: negative working capital plus a measurable performance-marketing engine is a machine, but the distribution is rented, so watch who controls the traffic.

The 30,000-Foot View

Booking is a two-sided travel marketplace. On one side are hundreds of thousands of hotels, apartments, homeowners, car rental agencies, and restaurants that want to fill inventory. On the other side are travelers looking for a place to stay or a flight to catch. Booking sits in the middle, matches them, and takes a fee. It owns no real estate and carries no inventory risk, which is why the model throws off so much cash.

The way value gets created is classic marketplace liquidity. More properties on the platform give travelers more choice, which brings more travelers, which makes the platform more valuable to properties. That flywheel is real, and Booking spun it up better than almost anyone, especially in Europe and Asia where Booking.com and Agoda are the default. The company's stated strategy is the "Connected Trip," stitching accommodation, flights, cars, attractions, and dining into one integrated experience so a traveler books the whole trip in one place. Flights have been the fast-growing wedge here, and management has said Booking may now be one of the largest third-party sellers of flight tickets.

The catch is that a lot of demand still comes through paid channels, mostly Google. Booking is a performance-marketing machine that treats ad spend like a ROI-tested capital-allocation decision, not a fixed budget. That discipline is a genuine skill, but it also means a meaningful slice of gross profit gets handed to the traffic gatekeeper every year.

Revenue mix (FY2025):

• International (bookings processed through the Netherlands entity): ~89%

• United States: ~11%

• By activity, accommodations remain the large majority of gross bookings, with flights, rental cars, attractions, and OpenTable dining smaller but growing.

Key Stats

• Market cap: ~$135B

• TTM revenue: ~$27.7B (FY2025: $26.9B)

• TTM operating margin: ~34% (Booking reports revenue net of amounts paid to travel providers, so there is no traditional cost of goods and gross margin is not the metric to watch)

• 1Y total return: about -24%

• Employees: ~24,300

• Industry: Travel Services (Consumer Cyclical)

Company History

• 1997: Priceline.com is founded by Jay Walker, famous for its "Name Your Own Price" bidding model.

• 1999: Priceline goes public at the height of the dot-com boom.

• 2005: Priceline acquires Bookings B.V. (following Active Hotels in 2004), creating Booking.com and pivoting hard toward the European agency model. This is the single most important move in the company's history.

• 2007: Acquires Agoda, planting a flag in Asia.

• 2010: Acquires TravelJigsaw, which becomes Rentalcars.com.

• 2013: Acquires KAYAK for about $1.8B, adding metasearch and price comparison.

• 2014: Acquires OpenTable for about $2.6B, moving into restaurant reservations.

• 2018: Renames itself from The Priceline Group to Booking Holdings and changes its ticker to BKNG, reflecting that Booking.com had become the crown jewel.

• 2024: Initiates its first quarterly dividend and rolls out an AI Trip Planner powered by generative AI, leaning into the Connected Trip.

• 2025: Posts record revenue of $26.9B with operating margin near 34%, expands the Genius loyalty program beyond accommodations, and debuts agentic AI features.

Show Me the Money

Standout financial features:

• Revenue grew from $21.4B (FY2023) to $26.9B (FY2025), and operating profit nearly doubled from $5.8B to $9.3B over the same window. This is the definition of operating leverage!

• Building on the point above, pperating margin expanded from 27.3% to 34.5% in two years. On an asset-light model with almost no capex (about $322M last year against $9.4B of operating cash flow), that margin drops almost straight to free cash flow.

• Net debt is a modest $2.1B against roughly $17B of cash on hand, so the balance sheet has plenty of firepower even after $6.4B of buybacks and $1.2B of dividends in FY2025.

A quick note on the table below: because Booking books commission and agency revenue net of what it pays travel providers, its income statement carries no meaningful cost of revenue line, so reported gross margin is effectively 100%. The number that actually matters here is operating margin.

Financial Data

Stock Performance

Dividend-adjusted total returns for BKNG:

One-year total return versus two public online-travel peers:

The long run and the short run tell opposite stories. Over five and ten years Booking crushed it (roughly doubling in five years and more than tripling in ten), but the last year has been ugly, with the stock down about 24% while Expedia ran up about 35%. The gap isn't really about this year's numbers, which were fine. It's the market pricing in the risk that AI assistants change how people find and book travel. When a business grows earnings and the stock still falls, the story has moved from fundamentals to narrative.

The N.O.O.B. Nine: Competitive Powers

The Nerd Out on Business Nine is made up of Hamliton Helmer's famous "7 Powers" of competitive advantage (Scale Economies, Network Economies, Counter-Positioning, Switching Costs, Branding, Cornered Resource, and Process Power) combined with two of my own (Data Flywheel and Distribution Advantage).

Average Score: 3.2/5 - A real moat built on network effects, brand, and scale, held back by low switching costs and a distribution channel Booking rents rather than owns.

Memorable Marketing

Booking's marketing has two very different faces. Underneath is a relentless performance-marketing machine that bids on search and metasearch at enormous scale and tests spend against ROI constantly. On top is a brand layer meant to pull travelers straight to the app so Booking pays less rent to Google over time.

Notable campaigns and tactics:

• "Booking.yeah" (ongoing since 2013): The long-running brand platform built around the feeling of getting travel right, designed to make Booking.com a name people type in directly.

• Celebrity brand pushes (2022 to 2023): Splashy campaigns featuring stars like Idris Elba and Melissa McCarthy to build top-of-funnel awareness in the US, where Booking is under-indexed versus Europe.

• Genius loyalty program: A tiered rewards program that converts expensive paid traffic into cheaper, repeat direct bookings, now expanding beyond hotels into rental cars, flights, and attractions.

• Performance-marketing engine (always on): The real workhorse, treating every marketing dollar as a measured investment rather than a fixed budget.

Tactical takeaways:

1. Treat marketing as capital allocation with a hurdle rate, not a fixed line item. Booking will spend more when the ROI is there and pull back when it isn't.

2. Use a loyalty program to turn your most expensive channel (paid acquisition) into your cheapest one (direct repeat).

3. Layer brand advertising on top of performance marketing to lower blended customer acquisition cost over time.

4. Own the app relationship, because the home screen is the one piece of distribution you don't have to rent.

AI Uses & Opportunities

Current exposure:

• Booking has shipped generative AI tools including an AI Trip Planner (built with OpenAI's models) and Priceline's "Penny" assistant, and has integrated with third-party assistants like ChatGPT for trip planning.

• Management frames AI as a discovery channel rather than a pure threat, positioning Booking to be the bookable, curated layer underneath whatever assistant a traveler starts with.

Future opportunities:

• Agentic booking, where an AI agent handles the entire Connected Trip (hotel, flight, car, dinner) end to end on the traveler's behalf.

• Cross-selling OpenTable dining against Booking.com traveler data, an underused combination management has flagged.

• AI-driven customer service and re-booking automation to cut support costs across hundreds of millions of trips.

• Dynamic merchandising and personalized pricing powered by the booking data flywheel.

Once again, If ChatGPT, Claude or Google's assistant becomes where trips get planned and booked, Booking could be pushed one layer down the stack, from the brand travelers choose to a commodity supplier the agent picks. That is exactly the fear the stock is pricing in.

Bumps in the Road

• Rented distribution. A large share of traffic flows through Google, which keeps raising the cost of that channel while building competing travel products of its own.

• AI disintermediation. Agentic assistants could insert themselves between Booking and the traveler and erode the direct relationship.

• Low switching costs. Travelers and hotels multi-home freely, so there's little lock-in on either side of the marketplace.

• Regulation. Booking is designated a gatekeeper under the EU's Digital Markets Act and has faced years of European scrutiny over rate-parity clauses, which constrains how it can operate.

• Cyclicality and FX. Travel is discretionary and roughly 89% of revenue is international, so a downturn or a strong dollar hits reported results.

Your Swipe File

• Booking is a case study for the power of two financial levers: negative working capital and operating leverage.

• Buy the category leader in each vertical instead of building it. Booking.com, KAYAK, OpenTable, and Agoda were all acquisitions that became defaults, not internal projects.

• A loyalty program is a channel-arbitrage tool. Genius exists to convert costly paid clicks into cheap direct repeats. Loyalty programs are underutilized in many small business niches.

• Watch who owns your distribution. Booking's biggest structural weakness is that its largest demand channel is one it rents from a competitor. If your growth depends on a platform you don't control, that is a tough place to operate. Try like hell to diversify!