Sam Shank on AI’s Challenge to Online Travel Giants and What’s Next

The online travel industry may be approaching a real distribution shift, and Sam Shank, co-founder and former CEO of HotelTonight, framed the pressure point clearly at the Skift Global Forum on October 20, 2026. In a Good Morning Hospitality bonus episode, Shank argued that AI-powered booking agents could challenge the sponsored search model that has long supported major online travel agencies such as Booking.com and Expedia.

That matters because online travel has not just been a story about who has the best inventory. It has also been a story about who controls discovery. If AI agents become the first place travelers express intent — not a search box, but a conversational assistant that understands budget, timing, loyalty preferences, neighborhood tradeoffs, and trip purpose — then the economics of travel distribution could start to move away from paid placement and toward machine-readable inventory, direct integrations, and specialized service layers.

Why AI agents matter beyond convenience

The easy version of the AI travel story is speed: fewer tabs, faster comparisons, less manual filtering. Shank’s point is more disruptive than that. An AI booking agent does not have to behave like a traditional OTA storefront. It can act more like a decision layer sitting between the traveler and the supplier.

That changes the value chain. Traditional OTAs have benefited from being the place where consumers browse, compare, and click. Sponsored search and paid ranking have been central to that model. But an AI agent can gather options in the background and present a recommendation based on user intent rather than visible ad placement.

If that behavior becomes normal, the question for the industry is no longer just who has traffic. It becomes who has the best data connections, the cleanest booking rails, and the strongest ability to serve an an AI-mediated customer journey.

For travelers, that could mean a booking experience that feels less like shopping a digital shelf and more like delegating a task. For suppliers, it could mean a chance to reach demand without paying the same kind of discovery toll to a dominant intermediary.

The sponsored-search model is the pressure point

Shank’s critique lands on one of the most important mechanics in online travel: sponsored search. Large OTAs have built powerful businesses in part because they aggregate demand and monetize visibility. Hotels and other suppliers compete for placement, and that placement influences what consumers see first.

AI agents threaten that logic because they can hide the ranking layer from the user. If a traveler asks for a quiet hotel near a meeting venue, with flexible cancellation and a strong value proposition, the agent may return a short list without exposing the paid-placement contest that has historically shaped results pages.

That does not automatically eliminate monetization. It does, however, force a rethink. OTAs may need to adapt from selling visibility on a page to competing on fulfillment, trust, packaging, loyalty, service, and infrastructure. In other words, the interface advantage could weaken if the interface itself becomes conversational and portable.

There is also a consumer consequence here. In theory, AI-led discovery could improve relevance and transparency. In practice, it raises a new question: whose incentives shape the recommendation engine? A sponsored listing on a classic OTA is at least a familiar concept. A recommendation generated by an AI agent may feel neutral even when commercial incentives still exist behind the scenes. That means the next battle may not be over whether ranking is paid, but whether the economics of recommendation are disclosed clearly enough for users to trust them.

Who could gain if the market opens up

One of the more interesting parts of Shank’s argument is that disruption does not only threaten incumbents; it can also create room for narrower, more focused winners.

Specialized agents could emerge around specific use cases rather than trying to replicate the full OTA model. Last-minute stays, premium travel, sustainability-focused itineraries, group coordination, or business travel workflows are all examples of areas where a more tailored AI layer could outperform a general marketplace. HotelTonight itself was built around a focused use case, so Shank’s view carries some weight here: specialization can matter when the product experience is sharper than the broad platform alternative.

The infrastructure side may be just as important. If AI agents need reliable access to rates, availability, policies, and booking confirmation, then the companies that provide APIs, direct connections, and transaction rails could become more valuable. In that world, the winners are not only consumer-facing brands. They may also be the less visible firms that make AI booking possible at scale.

For hotels and travel providers, this is the strategic opening. If they can expose inventory and policies in ways AI systems can use cleanly, they may gain leverage relative to platforms that historically controlled demand. But that opportunity comes with work: better integrations, cleaner content, clearer cancellation terms, and more dependable fulfillment.

Why the timing matters now

This debate is happening in a macro environment that is not especially forgiving. Recent market moves show investors are still repricing the balance between near-term policy and longer-run growth expectations. As of October 19, 2026, the 2-year Treasury yield moved 3 basis points while the 10-year yield moved 6 basis points, a split that suggests the market is adjusting its view of policy and growth at the same time rather than settling into a simple risk-on narrative.

That backdrop matters for travel technology. New booking models need capital, supplier adoption, and consumer trust. Higher yields and a more selective funding environment can make it harder for challengers to scale quickly, even when the product thesis is compelling. At the same time, pressure on margins can make suppliers more willing to test alternatives that promise lower distribution costs.

The broader US data context is mixed rather than one-directional. CPI rose to 336.298 in September from 334.131 in August, while the effective federal funds rate stood at 3.75 in September. Unemployment was 4.2 in September. Retail sales increased to 745021.0 from 739322.0, but University of Michigan consumer sentiment was 51.7 in August, down from 55.2. That combination matters for travel: consumers are still spending, but confidence is not especially strong, which can make value, personalization, and price transparency more important in booking decisions.

The tradeoffs and caveats

The bullish case for AI travel agents is straightforward: better matching, less friction, more direct supplier access, and less dependence on paid placement. But there are real constraints.

First, trust is hard to win in travel because mistakes are expensive. A weak recommendation in shopping is annoying; a bad hotel booking, missed policy detail, or failed itinerary handoff can ruin a trip. That means AI agents will need to prove not just intelligence, but reliability.

Second, incumbents are not standing still. Booking.com and Expedia have scale, brand recognition, supplier relationships, and large user bases. Even if AI changes the interface, those companies can invest, partner, acquire, or adapt their own products. Shank’s thesis is disruptive, but disruption in travel often takes longer than the technology narrative suggests because the industry is operationally complex.

Third, suppliers may want more direct access to demand, but they also value the volume and reach that large OTAs provide. The likely outcome is not an overnight replacement of the old model. It is a more contested middle ground where AI agents, direct channels, and incumbent platforms all compete to own the customer relationship.

What travelers and industry players should watch

For travelers, the practical question is whether AI tools start producing meaningfully better booking outcomes, not just prettier interfaces. Do they surface better-fit options? Do they explain tradeoffs clearly? Do they handle changes and cancellations well? Those are the tests that matter.

For hotels, airlines, and other suppliers, the watch point is whether AI agents begin to control enough demand to justify dedicated integrations and commercial terms. If that happens, distribution strategy could shift from optimizing OTA placement to optimizing machine-readable access and conversion quality.

For investors and operators, the key signal is whether AI becomes a true discovery layer rather than a feature bolted onto existing storefronts. If the answer is yes, the economics of travel search, ranking, and commissions could change more than many incumbents would like.

That is why Shank’s comments are worth paying attention to now. The core issue is not whether AI can help plan a trip. It is whether AI becomes the gatekeeper between traveler intent and supplier inventory. If it does, the online travel giants will face a deeper challenge than a new user interface.

For readers interested in the evolving travel and market landscape, broader context on pricing and consumer conditions can be useful. Readers comparing market access around this story can review Best Brokers, while those tracking inflation context can also use What is CPI.

Related reading

A useful background piece for this story is Best Brokers.

Readers who want the wider market context can also use What is CPI.

Sources

• GMH Bonus: Sam Shank on How AI Could Reshape Online Travel
• CPI — Consumer Price Index (FRED official data)
• PCE — Personal Consumption Expenditures Price Index (FRED official data)
• UNRATE — Unemployment Rate (FRED official data)
• PAYEMS — Nonfarm Payrolls (FRED official data)
• FEDFUNDS — Effective Federal Funds Rate (FRED official data)
• DGS10 — 10-Year Treasury Yield (FRED official data)
• DGS2 — 2-Year Treasury Yield (FRED official data)