TravelAI has given Sonder.com a second life, but not as the company travelers used to book. The Vancouver firm bought the brand out of bankruptcy, left the leases behind, and turned the name into an AI-enabled guide for urban stays.
Sonder is back. Sort of. The rooms are not. TravelAI, the Vancouver-based company operating under UpNext Ventures, announced on July 27 that it had acquired the Sonder word mark, more than 50 global trademark registrations and more than 70 domain names, including Sonder.com. The old hospitality business is still gone. TravelAI didn't buy the leases, buildings, reservation records, staff or inventory, and the company's investor page says the Quebec court approved the sale on July 6 before the transaction closed on July 13.
That distinction is the story. You can still type Sonder.com into a browser and find a familiar travel name, but the company holding the keys has changed completely. As Skift reported on July 27, TravelAI bought the brand carcass for its residual search traffic and recognition, not for the operating company that collapsed in late 2025. PhocusWire also reported that the terms of the deal were not disclosed.
TravelAI CEO John Lyotier said the quiet part plainly in the company's release. "Sonder earned something rare: a name people remember and a standard of taste they trusted," he said. He also said TravelAI "did not acquire a portfolio of buildings." The new Sonder is being pitched as a curated guide to apartment-style stays, boutique hotels and urban accommodations, with bookings fulfilled through TravelAI's partners rather than through a resurrected Sonder operating platform.
That's the asset. Not the front desk. Not the lock on the door. TravelAI says its wider network now includes more than 530 travel brands serving more than 50 million travelers annually across 200 countries, and it wants Sonder to sit on top of its Traveler.md memory infrastructure, which is designed to carry a traveler's preferences from one trip or brand to the next with permission. You don't have to buy every claim about AI personalization to see the business logic here. A remembered domain is cheaper than a lease portfolio.
To understand why this move matters, you have to remember why Sonder failed. On paper, the model looked clever: lease apartments and boutique properties at scale, standardize the design and service, then sell the result as a more predictable alternative to short-term rentals. It sounded asset-light. It wasn't.
Sonder's own October 2025 filing showed the problem in hard numbers. For the second quarter of 2025, occupancy reached 86%, revenue per available room rose 13% year-over-year and adjusted free cash flow was still negative $17.5 million. The company could fill rooms and still burn cash. A long-term lease behaves like a hard obligation even when a pitch deck calls the model flexible.
Then it broke. Marriott announced on November 9, 2025, that its licensing agreement with Sonder was no longer in effect because of Sonder's default. PhocusWire reported the next day that Sonder planned to wind down operations immediately and initiate Chapter 7 liquidation of its US business, citing severe financial constraints and delays in integrating its systems and booking arrangements with Marriott. CNN Business later reported guests were stranded mid-stay across Sonder's 9,000 furnished apartments and boutique hotel rooms in 40 cities and 10 countries.
That part checks out. The collapse was not a slow brand fade. It was the kind of failure travelers remember because it interrupted real trips, not because it disappointed investors.
The name survived the model #
TravelAI's bet is that the Sonder name still carries value once you strip away the liabilities that sank the old company. Frankly, that's the cleanest version of the deal. You get the domain authority, brand recall and travel intent, but you don't inherit landlords, payroll or guests who already have bookings in a broken system. The new Sonder sends travelers toward partner inventory instead of trying to control the whole stay itself.
This is a model worth watching because distressed consumer travel brands now have a second buyer class. They don't have to be bought by hotel operators. They can be bought by distribution companies that want demand, search history and a name people already know. TravelAI has done this before - with Casai, Smartours and OwnerDirect, according to the company's release and BetaKit's coverage of the Sonder acquisition. That's the playbook.
Davidson starts over without the leases #
Francis Davidson, Sonder's co-founder and former CEO, is also trying to build the version of travel software Sonder never became. PhocusWire reported in June that his new company, Odessia, raised $6 million led by Sequoia to build a conversational AI travel concierge. The pitch is conversational: Davidson wrote on LinkedIn that the product can plan a trip to Mexico City, show flights, hotels and an itinerary, then make the selections bookable from the same flow.
No leases. No properties. No cities full of stranded guests if a licensing deal falls apart. Odessia is still early, and a demo is not a durable business, but the contrast with Sonder is sharp enough to matter. Davidson is staying in travel while walking away from the fixed real estate exposure that made the first company so brittle.
The real verdict on asset-light hospitality isn't that the idea was stupid. It is that Sonder confused no ownership with no obligations. A lease in Manhattan or London is not light just because the company doesn't own the building. TravelAI appears to have learned that lesson from the wreckage. Davidson appears to have learned it too. The next test is whether either company can turn that lesson into bookings without rebuilding the same risk under a cleaner name.
Also read: Cursor ships Moonshot AI's record 2.8-trillion-parameter Kimi K3 on launch day | OpenAI quietly slashed GPT-5.6 Sol's reasoning power by 87% four days after launch | Build Your Startup Hiring Plan Before the Runway Runs Out