An empty market usually has a reason.

The gap has causes. Each condition needed to fill it fails on its own, and they fail in a set order: supply, then control, then regulation, then economics. An owner who understands that order can still build something durable. An owner who treats the gap as an oversight will spend three years learning the order at their own expense.

The order leads to a question that reaches well beyond hospitality.

Everything else depends on how many buildings there are. Of twenty-five canonical works examined, about half are run as museums or by institutions that allow no overnight stays. Another group are privately held with no public access at all. The rest, around a dozen, can be stayed in today.

A dozen buildings is a fixed ceiling. They belong to owners with nothing in common: a state monument service, a Swiss family, a Japanese corporate foundation, a British preservation charity, an individual collector. Scarcity defines this asset class, and it explains why the category has no established operator.

Alain de Botton built Living Architecture, the only architecture-led rental business with published occupancy figures. In 2012 he was asked whether it could grow to the size of the Landmark Trust. He named the constraint exactly: “the reason it could expand so fast was that the properties were already there.”

The verdict. The buildings within reach are a dozen, held by owners with nothing in common, and no amount of capital creates a thirteenth.

Scarce supply would be manageable if the owners were willing. Owners of canonical architecture are a narrow set: trusts, states, museums, foundations and families. Each already has an answer to the question a network would ask.

A preservation charity lets its buildings to pay for their repair, and the repair is the purpose. The Landmark Trust’s aims are to preserve historic buildings and promote public enjoyment. Income from the two hundred buildings in its care supports their maintenance and survival. A family that has held a house for three generations has no use for a booking network. A state monument service can hand over control of pricing only through a procurement process.

An asset-light network, one that runs buildings without owning them, needs three things from these owners: control of brand, pricing and booking. None of the owners has a reason to give them. Where overnight stays do exist, they came about another way. A family business grew into a hotel, a collector opened their own house, or a charity converted a building to fund its repair.

The verdict. Gathering these buildings into one network fails on incentives, before capital even comes into it.

Suppose both problems were solved and an owner agreed. The rules for small heritage buildings then cap the business at a level most financial models never reach. Those models start from room rate and occupancy.

In France a chambre d’hôtes (bed and breakfast) may offer five rooms and host fifteen guests at once. Above either figure the designation cannot be used. The property then falls under the rules for establishments receiving the public, with the fire and accessibility duties that follow. In Japan, the law defines a private lodging business as one used for no more than 180 days a year. In London, temporary sleeping accommodation is allowed for fewer than ninety nights in a calendar year. Beyond that, the change of use needs planning permission.

Together the three show the pattern. One caps the rooms, one caps the nights, and one in effect caps both. A five-room house let for half the year is a different business from the one in the spreadsheet, and no nightly rate closes the gap. The limit hits revenue before the first guest arrives.

Tax makes it worse. In several countries, heritage tax relief is designed for buildings that are kept and shown to the public. Commercial use above modest thresholds removes it. An owner who converts often trades a tax position already held for revenue not yet proved.

The verdict. The building’s legal category sets the ceiling before the model starts, and quality cannot raise it.

Leave the buildings aside and look at how hospitality businesses of this kind make money.

The first model is owned real estate. When LVMH acquired Belmond in December 2018, it paid twenty-five dollars a share. That was an equity value of 2.6 billion dollars within an enterprise value of 3.2 billion. In the twelve months to September that year, Belmond recorded 572 million dollars of revenue and 140 million of adjusted EBITDA, a measure of operating profit. It had forty-six properties in twenty-four countries. The model works and earns a high multiple, but it runs on owned property at a scale that takes decades to assemble.

The second is the management fee. Two months later IHG bought Six Senses for 300 million dollars in cash, explicitly without real estate. The announcement put its fee revenues at more than thirteen million. The multiple is high because fee income is nearly all margin and needs no capital. It is open only to an operator with enough properties under contract for the fees to add up, and those properties belong to owners who agreed to pay them.

The third is stewardship. In the year to 31 December 2024, the Landmark Trust had income of £20,232,000 and expenditure of £20,410,000, across two hundred buildings. Sixty years of institutional trust and the best-known portfolio of its kind in Britain produce a small deficit. The National Trust is far larger and ends up in the same place. Its holiday cottages and campsites generated £22.5 million in 2024–25, while its total income was £766.2 million. In April 2026 it confirmed that more than 130 of those cottages would be repurposed after a review of financial sustainability.

The verdict. An owner who picks a model without its precondition gets its failure mode.

The demand case is usually the weakest part of the proposal, and the most confidently written. Two facts sit awkwardly together.

Architecture-led stays do sell. Living Architecture’s occupancy was “often close to 100 per cent on many of the houses, and never falls below 90 per cent”. The same RIBA Journal interview put the industry average at seventy per cent or lower. That is the strongest single demand signal in the field. It is also more than a decade old, based on five houses, and comes from the founder.

The guests behind that occupancy book today through channels priced in the low hundreds per night. Guests able to pay luxury rates have shown they prefer owning trophy property to renting someone else’s. Nobody has credibly sized architecture tourism as a category, and the figures in circulation come from vendors whose methods do not hold up.

This leads to another temptation. If the audience is real but thin, perhaps the business is media, with stays as the proof. Airbnb tested this at a scale no cultural owner can match. Its Icons programme, launched in May 2024, placed guests in extraordinary buildings. At the time it was reported as free or low cost by design, meant to work as an alternative to advertising. The stays were the campaign.

The verdict. Content can set apart a business that already earns. Nobody has shown that it can create one.

Most of these conversations open with one question: can a network of architecturally significant houses be assembled and scaled? The answer is no. The reasons are structural, and a better pitch deck will not change them.

The useful question is the one underneath it.

Owners who answer these three in order tend to build something smaller than their first plan, and far more durable. Owners who skip to the third build the plan first and discover the other two later.