The $8.6 Trillion Payments Gap in Travel and Transportation

Stefan Sandoval

May 14, 2026

•

9 min.

As you’re reading this, a freight broker in São Paulo is confirming a load in a WhatsApp group for an invoice that won't settle for 47 days, a hotel clerk in Bangkok is hand-keying a credit card number into their property management system that hasn't had a UI refresh since 2007, and a passenger at JFK is paying for a checked bag through the customer service kiosk, where their money is routed through five intermediaries before the airline ever sees a dollar.

This is just an average day in travel and transportation payments — one of the most fragmented, underserved corners of fintech.

Travel and transportation move serious money, but the technology hasn’t kept pace. To put it in perspective, global travel spend hit $8.6T in 2024, accounting for ~ 9% of GDP. Take our São Paulo freight broker as a micro case study: the freight factoring market is forecast to nearly double from $200B in 2026 to $437B by 2035.

So, why are $20K travel invoices being processed by an emailed PDF, while $50 ecommerce orders from big sellers like Amazon and Walmart have been refined and optimized for decades? It’s because travel and transportation payments are too fragmented for any single player to fix, with the sector split across airlines, hotels, freight brokers, OTAs, GDS systems, and fuel networks, each with its own intermediaries and settlement logic.

Visa, arguably one of the largest payment platforms in the world, predicted that by 2028, AI will deliver the "perfectly tailored itinerary" with booking, payment, ancillaries, and contingencies stitched together by a single interface on the traveler's side. While we agree that this is where the industry is headed, the timeline doesn’t account for how broken the current system is, and the connected trip that Visa is pitching assumes infrastructure that doesn't actually exist yet. The real opportunity is that AI-native tooling finally lets builders do this cleanly, augmenting legacy rails where they don’t need to be torn out, and seamlessly replacing them where they can be, one vertical at a time.

In this piece, we focus on airlines, hotels, and trucking because together they cover the B2C-to-B2B spectrum of the sector.

Aviation

Airlines sell ancillary upgrades to nearly every passenger: baggage, seat selection, priority boarding, and in-flight purchases stacked on top of the base ticket. The industry has layered buy-now-pay-later, modern booking APIs, and fraud detection on top of traditional card infrastructure.

A single ticket purchase for an airline can pass through five different points – a customer's bank, a payment gateway, a merchant bank, a card network, and an orchestrator – all before the carrier sees its share. In this space, some of the most interesting work on modernizing payments is happening at the orchestration layer.

Duffel has built a single API to 500+ airlines across 40 countries and does $900M+ in total transaction volume.
CellPoint Digital has signed Riyadh Air and Southwest as anchor customers for an orchestration platform that uses ML for routing, fraud, and acquirer decisions on every transaction.
Hands In is going after a different leak: card declines driven by issuer credit limits, which quietly cost airlines and OTAs meaningful revenue. Their split-payment technology lets a single transaction spread across multiple cards or travelers, recovering purchases that would otherwise fail.

The opportunity in aviation sits in the seams between the carrier and the legacy stack. Orchestration is the wedge that's already opened, and early movers there have shown vertical AI rails can win. Fraud, for example, is another space with tremendous opportunity. With travel fraud projected to hit $362B by 2028 and chargebacks growing roughly ~30% a year, the risk and dispute layer needs innovation. One of our portfolio companies, UrbanFox, is already deployed at major airlines and takes a different approach. Their solution uses proprietary generative AI and behavioral models to catch fraud earlier in the transaction flow, before it reaches payment processors, without relying on static rule sets that fraudsters have learned to beat.

Hotels

A hotel’s main pain point lies in its cash flow. Online travel agencies (OTAs) like Expedia or Booking.com drive about $40B in bookings, roughly a quarter of the global hotel market, and charge 15-25% commissions on each sale. They collect the guest's payment at the time of initial booking, but the hotel doesn't see that money until the customer actually checks out, sometimes weeks or months later.

Two angles of attack are emerging: bypass the OTAs entirely, or modernize the rails hotels already run on. Selfbook, as an example, is attacking the direct-booking side with one-click hotel checkout. Mews, the hospitality operating system with embedded payments, processed $19.7B in transactions in 2025, and can be seen as a categorical winner.

Short-term rentals like Airbnb and VRBO have already gotten this right. The company owns the booking, the payment, and the fraud risk end-to-end, with no OTA taking a cut and nothing standing between the guest paying and the host getting paid within a business day of check-in. The fraud problem is its own animal too, focused on verifying new hosts and guests rather than fighting chargebacks on past stays .

The opportunity in hotels sits in shrinking the cut online travel agencies take and closing the cash flow gap between guest payment and hotel settlement. Both wedges are already opening, with direct-booking platforms pulling demand back to hotel-owned channels and embedded-payments operating systems modernizing the rails on top of the legacy stack. The space is wide open for vertical AI tooling that lets hotels capture more of the booking, settle faster, and run modern risk models without becoming payment companies themselves.

Trucking & Logistics

Trucking and logistics are mostly a B2B story of fuel cards, freight invoices, and driver disbursements. Despite the scale of the industry, a surprising amount of the workflow of moving products from A to B still runs on email and group chats, which is exactly the gap startups like OTIF and Nauta are stepping in to close.

Trucking has two payment problems running in parallel:

Fuel cards clear within days, while freight invoices (bills sent after delivery) can go unpaid for weeks.
Drivers are almost always last in line, paid only after fuel and freight invoices are clear. The gap that still surprises us: 97% of logistics companies pay fuel on time, but fewer than half pay drivers with the same urgency, even though 93% of drivers want instant disbursement.

Of the three verticals, this is the cleanest operator-level payments problem and the largest under-the-radar AI opportunity. Most of the existing activity sits at the fuel-card layer, where the incumbents are entrenched – the OTR fuel-card market is dominated by Wex and Corpay, with 99% of US trucks accepting one or both. Series C startup AtoB is moving in as a third player, partnering with Uber Freight and acquiring India-based LogiPe to expand globally.

The AI opportunity is less glamorous than agentic checkout but probably more economically significant. Multi-agent systems that handle invoice generation, payment tracking, and dispute resolution can eliminate a meaningful share of billing discrepancies.

The trust bottleneck

Across the three verticals, the same pattern keeps surfacing: the technical layer is moving faster than the trust between counterparties. The agentic future Visa described requires AI agents that can negotiate, settle, and reconcile on behalf of operators who don't fully trust each other yet, and that trust took decades to encode in the rails the industry currently runs on.

Two airlines won't trust each other's agents to interline ticket changes. A 3PL won't trust a broker's AI agent to release payment on its behalf. A hotel chain won't trust an OTA's agent to confirm group-rate inventory at the moment of booking. Each is a small instance of a larger structural reality. Trust between counterparties in this sector is slow to renegotiate, and it won't be replaced by generative AI overnight.

The technical preconditions are catching up. Generative AI is already deployed in narrower use cases such as chargeback automation, refund eligibility, and fraud screening, mirroring where e-commerce stood a decade ago. The trust scaffolding will eventually settle, but operators don't get to wait that long.

Build, outsource, or wait for the agents to talk

For operators whose core business isn't payments, building it in-house is rarely the right call. Obvious as that may sound, the question still comes up in most conversations with founders, investors, and LPs today.

The cleanest play is to outsource the payment infrastructure/pipes and own the experience. An operator that can’t build a high-uptime payment infrastructure also won’t be able to build a high-quality AI fraud model, dispute engine, or personalization layer comparable to what Outpayce, Riskified, Forter, and Sift offer as a service. Operators are better off doing what they do best: owning the customer relationship (and the data feeding it) without being bogged down by trying to create a ML model or routing engine. The space this creates for emerging startups is vertical-specific AI rails: flexible, well-supported, and aimed at the work operators won't try to rebuild themselves. , and

Where we're focused

Travel and transportation payments will be rebuilt industry by industry over the next five years, with AI as the engine and trust between counterparties as the gating factor. Operators that win the next phase will pick the right battles to build in-house and outsource the rest. For startups, the moats that survive incumbent counterattacks are built on operator-specific solutions, vertical-specific rails, or loyalty data.

We're watching the operators who're getting build-vs-buy decisions right, the startups making credible plays on vertical-specific rails, and the brands figuring out how to be an AI agent's default.

If you're building in space, we'd love to talk. Reach out to us at inbound@sky-vc.com.