Travel 2045: A 20-Year Outlook for the AI Era

Every January, the travel industry is flooded with “trend outlooks.” Reports, predictions, and glossy lists of what might shape the next 12 months. Some are useful (at least for inspiration). Many simply repackage shifts that have been underway for years. Almost all are too short-sighted.

Because if the goal is to prepare for the future of travel, then a single year is the wrong unit of analysis.

The forces reshaping our industry don’t operate on annual cycles. They unfold over decades through deep structural shifts in technology, traveler behavior, labor dynamics, and the digital infrastructure that powers global travel. You don’t prepare an airline, airport, or OTA for next year. You prepare it for the next era.

That’s why this outlook takes a different approach.

Instead of asking “What will happen in 2026?”, we ask:

“Where will travel be in 20 years?”

Such a horizon forces us to zoom out.

It shifts our attention away from year-to-year fluctuations and toward the deeper forces that reshape industries over decades. We are talking about the kinds of shifts that determine which travel companies lead, which fall behind, and which new categories may emerge altogether.

This report is built around ten long-term bets that will shape the next two decades of travel technology.

At OAG, we believe these next 20 years represent a once-in-a-generation shift, powered by the rise of agentic travel. We’ll get to that shortly.

But before looking forward, we need to look back.

Because the past 20 years have already transformed travel in ways that are easy to forget and impossible to ignore.

To understand where the industry is going, we must first understand how it arrived here.

So we begin with five defining shifts that shaped travel between 2005 and 2025, and then turn to the ten bets that will shape the 20 years ahead.

A simple but powerful truth about human behavior: we optimize for maximum output with minimum effort.

And nowhere is this clearer than in travel planning.

Research shows just how overwhelming the trip-research phase has become.

• Expedia Group found that travelers spend an average of 128 minutes on airline websites during the path to purchase (so more than two hours hunting for fares).
• While a handful of travelers may enjoy the chase, for most it’s pure frustration. In fact, more than three-quarters of travelers say comparing airline prices is too time-consuming.

And it doesn’t stop with flights.

Research suggests that travelers (on average) visited 141 web pages before booking a trip (flight + hotel + extras) in 2024, up from just 38 in 2013. In the U.S., that number spikes to 277 pages per trip.

This burden will be increasingly outsourced to AI.

It’s also the primary reason why more than 50% of travelers already use ChatGPT or similar tools during their trip planning (as mentioned above). People don’t want to do the repetitive work, and AI is finally capable of doing it for them.

Travel isn’t only about efficiency. At its core, it’s about discovery and seeing something new, unexpected, or unplanned. Yes, most travelers still visit the famous places they’ve researched in advance. But almost everyone also craves a moment off the beaten track: a hidden café, an unexpected view, a local ritual you didn’t have on the itinerary.

This might look counterintuitive, but AI agents will actually add more serendipity, not less.

The first wave of agentic systems focuses on efficiency (as we already see today in the perfectly crafted travel itinerary), aiming to reduce friction, cut research time, and optimize routes. But as these systems mature, the next frontier will be curated surprise. AI won’t just echo back what we already like or what we explicitly asked for. It will start to introduce small, well-judged deviations from our patterns.

A simple example:

If a traveler has a free afternoon in Tokyo, the agent might propose:

“You said you love modern art, but there’s a tiny pottery studio two streets away holding an open workshop today. Highly rated by locals. Takes 45 minutes. Want me to slot it in before your dinner reservation?”

How exactly this will be implemented remains to be seen…

Travel supply has always been, and will always remain, deeply fragmented.

Hundreds of airlines. Millions of hotels. Tens of millions of tours, activities, and restaurants. Plus loyalty programs, ancillaries, insurance products, rail systems, ferry networks, rental cars, local transit…the list keeps growing.

A trip will always consist of countless individual pieces owned by different suppliers.

And no amount of technology (not even AI) will consolidate this complexity away.

But here is what will change:

For the first time, AI can make fragmented supply feel unified.

• Historically, every layer of the trip required its own search, its own booking, its own confirmation, and its own support channels, meaning a mess of disconnected micro-decisions. 
• Even sophisticated travel platforms have struggled to connect these pieces “fluently.”

Agentic AI can finally operate across this fragmentation.

It can:

• Pull availability from multiple suppliers at once.
• Assemble multi-step itineraries dynamically.
• Detect hidden constraints between suppliers (timing, policies, location gaps).
• Optimize combinations that humans or traditional systems would not surface.
• Handle rebooking across multiple providers as one coherent action.
• Unify communication and service across disparate systems.

In other words: the supply side stays fragmented, but the traveler experience becomes more unified.

The industry doesn’t become simpler. It just becomes simpler to navigate.

While technology and AI accelerate at breathtaking speed, physical infrastructure will not.

Aircraft, airports, hotels, rail systems, and runways; none of these evolve on a 5-year or even 10-year cycle.

• The aircraft we will fly in 20 years have largely already been ordered today.
• Most airports expanding capacity now will finish construction late in the 2030s.
• Hotel pipelines move slowly, with lead times of 5–10 years even for mid-sized developments.

Infrastructure simply cannot keep pace with demand. And it never will. The timelines don’t move at the same speed.

So the only way to unlock more capacity is not by building more infrastructure, but by managing existing infrastructure far more intelligently.

This is where AI steps in. In the airline operations context, we already see countless use cases where AI takes planning and efficiency to a new level (see our collaboration report with Microsoft). 

The next 20 years will push this much further.

AI will help decision-makers answer questions such as:

• Where should aircraft be positioned to minimize disruptions?
• Which gates should be reassigned to optimize flows and reduce delays?
• How can turnaround processes be improved in real time?
• Which crew rotations minimize risk under irregular operations?
• How do we predict bottlenecks hours before they materialize?

And underlying all of this is one foundational requirement: trusted, high-quality data.

Without reliable schedules, gate information, flight status, delay patterns, airport capacity data, and airline operational inputs, none of these AI systems can function.

So infrastructure won’t accelerate. Intelligence will.

Over-tourism is one of travel’s biggest structural challenges. Today, 70% of global travelers are believed to visit just 4% of all destinations, creating overcrowded hotspots while countless equally remarkable places remain overlooked.

This imbalance won’t fix itself.

And traditional marketing, recommendation engines, or city-level incentives have never truly shifted global demand.

But AI agents will change this dynamic.

Because agents won’t only optimize for personal preference or convenience, they will increasingly optimize for opportunity.

As they gain a more complete understanding of global supply, real-time demand patterns, congestion levels, pricing dynamics, and traveler intent, agents will have the power to:

• Surface under-visited but high-quality destinations.
• Redirect travelers to places that match their interests without the crowding.
• Time-shift demand away from peak periods.
• Propose alternatives when a preferred hotspot is overloaded or overpriced.

In other words, AI will act as a new kind of invisible hand, gently steering travelers toward the world’s hidden gems (not through restrictions or guilt, but through better matches).

Automation and agentic systems will take over large parts of the travel workflow, but travelers will only outsource tasks if they trust the provider, whether it's the payment platform, the booking environment, or most importantly, the agent itself.

Which agents will we trust? The ones that consistently prove their reliability over time. And the margin for error is almost zero.

• If an agent books the wrong flight, misunderstands a constraint, or mishandles a disruption, trust evaporates instantly. 
• Travelers won’t give second chances to systems that get core travel decisions wrong.

So which agents will earn (and keep) traveler trust?

Those with deeper context, better training, and, crucially, the best data to work with.

This brings us back to one essential truth for the AI era: trusted data becomes the foundation of trusted agents.

In our recent report on aviation’s AI future, we found that data inaccuracies are the number-one concern among companies running AI projects today. The industry knows it: without accurate, high-integrity data, AI systems cannot reason reliably, cannot personalize safely, and cannot make decisions travelers can depend on.

The key point: Travelers will continue to choose providers they trust. And in the AI era, trust will be earned through data quality.

If the past 20 years have taught us anything, it’s this: the breakthroughs that reshaped travel rarely came from within the industry. It was outsiders and early believers who recognized what the online era could mean:

• They created the first online booking and price comparison sites.
• They foresaw how critical seamless mobile payments would become.
• They reimagined entire travel categories that didn’t exist before, from home-sharing and ride-hailing to virtual interlining and experiential marketplaces.

But the next 20 years may look very different.

In the AI super-cycle, incumbents are structurally positioned to outperform.

Why? Because the success of agentic systems won’t depend on scrappiness or intuition, it will depend on scale, data, integration, and distribution.

Incumbents already have:

• Massive datasets spanning inventory, behavior, pricing, and operations.
• Aggregation layers across flights, hotels, mobility, and payments.
• Global distribution and loyalty ecosystems.
• The financial and organizational capacity to deploy AI at scale
• Trust and brand recognition that accelerate adoption.

Also, AI is becoming a horizontal capability and a commodity layer accessible to everyone.

It’s no longer the secret advantage of two brilliant founders in a garage, quietly building something no one else can. The “surprise moment” that powered past startup waves becomes far less potent when:

• The core technology is broadly available,
• Models and tools are open-access, and
• Implementation (not invention) becomes the differentiator.

This strengthens the hand of incumbents because the prerequisites for high-impact AI are already sitting inside the big players.

This doesn’t mean startups won’t innovate. Many will. But the companies best positioned to reimagine AI-powered travel at industry scale are the ones that already have the infrastructure, the data, and the distribution for AI to amplify.

Believers will still be needed.

But in this cycle, many of those believers will sit inside the incumbents, not outside them.