Big Interview: Travolution sits down with Oppenheimer & Co

Could you give us an overview of Oppenheimer & Co, what your role is there and why you’ve been given the closing keynote at Travolution Connects?

Oppenheimer is a publicly-traded company with 2,900+ employees across multiple different divisions – wealth and asset management, sales & trading, equity research, and investment banking, which is where I come in. There’s around 200 investment bankers, primarily across the US and Europe, with 40 focusing on technology. For the past nearly eight years, I’ve led our global travel technology investment banking efforts in partnership with the European team.

We advise small or mid-sized companies on selling their business or raising private capital. We also work with larger strategics looking to acquire growth-stage companies by advising on the buyside. Additionally, we’re very active in working with travel and transportation technology companies on their IPOs, including Navan, StubHub, and Via in 2025.

I get to know players from different sectors of travel at all stages of their development, so I am very much in the weeds of what is happening in travel tech.

So let’s get into the weeds! I read a LinkedIn article you posted recently where you highlighted “AI as an infrastructure layer” as one of themes on your radar – what do you mean by that?

We’re viewing the infrastructure layer as businesses who are providing the technology that agents and online distribution organizations need to sell travel. Most of the businesses we are advising today are B2B2C.

It seems like the place right now that’s giving the best risk-reward ratio is providing supply and the collateral you need for distribution, but not actually being the point of distribution, because much of that distribution sits with agents or incumbent online platforms.

Now we have this newer AI distribution point, which is fairly early in its development, but what hasn't been reported as much as it should be is that actual distribution through AI platforms has not widely taken hold yet.

I thought it was really meaningful two months ago when OpenAI showed that they are not going to get into the travel distribution business and compete with Expedia and Booking.

Why do you think that was?

I think it’s based around the complexity. There's a lot you need to build in. When there are changes, cancellations, or disruptions, there is a lot of customer service time involved. It can certainly be helped by AI though. Additionally, most people don't feel comfortable at this point dropping $5,000, $10,000 through an AI platform. How is it that we can be only a few years into mainstream adoption of AI and expect most people to be using it to book?

These are not brands consumers have come to trust over years and years, like an Expedia or a Booking.  We're 25-plus years into OTAs and still a material volume of bookings are made offline.

Is there a niche within travel as we know it where AI can be a gamechanger?

I think dynamic packaging is the part of the infrastructure layer where AI can play a very significant role, because dynamic packaging is complex. You bring together multiple products from different sources, and those products must also fit together in a way that makes sense for the traveler.

Ok, so you might be thinking, haven't Expedia and Booking been bundling or packaging forever? But when you get into flight plus hotel plus car rental plus activities plus maybe cruise…then you're starting to get really complicated. There's a lot of things for an agent to think through. The same goes for a dynamic packaging engine on an OTA’s website. We know where a lot of the margin often comes from…hotel, right? Packaging is a way to sell high margin products while also layering on incremental margin through a fully connected trip.

Margins are high in cruise as well, and cruise is still mostly booked offline, over the phone, with an agent. When someone books a cruise, they're also often booking a flight before, a hotel, a car rental to go around, and activities in the departure city. Then after the cruise they may stay in destination for a bit…which means another rental car, another hotel, another flight, another activity.

You may have several components to this booking outside of the cruise itself, which has its own layers of complexity. It’s a really good example of a travel booking which is difficult to dynamically package and can benefit from AI. 

There are many use cases for dynamic packaging. A fairly standard trip to, say, California can be put together in seconds because it’s quite contained…maybe a hotel in LA for two nights, car, Hollywood tour…then off to San Francisco, San Diego, or wine country. AI-enabled dynamic packaging technology can compile this whole thing with all these different pieces so quickly and can customize and re-price in real-time.

But isn’t this what travel agents have been doing for decades?

There’s a sweet spot in providing agents with an AI-enabled dynamic packaging engine that can create itineraries more quickly. The agent can then take the itinerary through to booking because of the trust there is between the agent and the traveler.

So if I’ve developed an AI-enabled dynamic packaging engine, what should I do?

I think the go-to-market is the whole battle here.  You’d have to get an agent to say, okay, I trust your technology enough for me to use it with clients I have had for a long time. The trust that clients have is an agent’s currency so they’re going to be hyper-protective of it. It is important to get proof-points with early adopters in a live environment to show that your engine has been tested and then pursue partnerships with large agencies and travel brands.

What are you looking for in a startup today?

We see a very product-driven environment today. At the moment it’s getting easier to build something – five years ago a startup often needed the budget for many developers and now a Claude subscription can give you a big head start. So the market is crowded with an excess of ideas vying for a limited amount of capital.

To get attention, startups need to show what they always have had to – MVP, go to market, customers…but there’s an added intensity at the moment to show a product’s enduring quality advantage and path to profitability.  Everyone is focused on invincibility against AI – how is this product not something that will get replicated?

Even if revenues are strong, to us product is more important. When we’re selling a business, we engage deeply with the product team so that we can effectively communicate the product’s strengths to the buyers’ corporate development teams. Even if the revenue growth and other financial metrics are very strong, we're in a unique place today where your business may be doing great, but everybody's really focused on how it may be replaced tomorrow.

What are you seeing when it comes to valuations and exits?

Exits have not been as prolific in this space and in general over in the past five years. 2021 obviously reset people's valuation expectations, and one important element of where we are today is that many businesses are stuck on valuation. This is behind the lack of exits, and this lack of exits has reduced exit preparedness.

Sometimes I’ve come out of conversations with investors being surprised by this lack of that preparation. Why can't we sell to Google? Because they’re not interested in a business that doesn’t move the needle.

And there are some businesses and investors aiming for valuations based on what the biggest and best are achieving, even if their growth profile and business model is different – you are not going to be valued like Booking or Airbnb when you’re doing $10 million or $20 million in revenue, lacking growth and burning capital.

Final thoughts?

Innovation across travel tech shows no signs of slowing down or tailing off – but many investors are in a holding pattern of sorts. While they are reluctant to jump in, or are making more considered decisions on where to invest, the desire remains strong to find and fund travel tech businesses with a solid profitability profile and AI defensibility.