Knight Frank’s luxury investment index closed 2025 down 0.4 percent, which the firm described as stabilisation after two years of decline. Inside that unremarkable number the categories pulled hard in opposite directions, and the widest gap was between two Swiss watchmakers.
The WatchCharts overall market index rose 5.1 percent for the year. Rolex’s index rose 4.6 percent. Patek Philippe’s rose 12.1 percent.
Same market, same twelve months, nearly triple the gain. The obvious explanations do not survive much scrutiny. Both houses make excellent watches. Both are more than a century old. Both have waiting lists. What separates them is a decision about volume.
One clarification is worth making before going further, because it is usually left out. These indices track what buyers pay each other, not what the houses charge. A watch that appreciates on the secondary market delivers nothing directly to the company that made it, which sold the piece once, at its own price, possibly years earlier. The number measures the strength of the promise attached to the object after it leaves the shop, which is a harder thing to manufacture than a price list.
The decision not to grow
Patek Philippe makes somewhere in the region of 60,000 watches a year. Rolex makes close to a million.
Patek could make more. It has the demand, the reputation and, one assumes, the capital. Thierry Stern, who runs the company, has been unusually direct about why he does not: production is kept where it is, and what exists is directed towards long-standing clients rather than expanded to meet whoever turns up. Waiting lists for the more sought-after references run for years.
This converts scarcity from an accident into an instrument. A house that cannot meet demand has a supply problem. A house that will not meet demand has a strategy, and the market prices the two differently, because the second one is a promise about the future. Anyone buying a Patek is buying the company’s continued refusal to print more of them.
The same instinct now shows up well outside watchmaking, in fashion houses that cap production runs and ateliers that quietly manage their client lists rather than their marketing budgets.
What else held its value
The rest of the index tells a consistent story about what survived a difficult year.
Fine art rose, with combined sales up 11 percent, led by categories where supply is fixed by the deaths of the artists. Fine wine kept falling; the Liv-ex Fine Wine 100 declined 2.5 percent, continuing a slide that began after the 2022 peak. Hermès Birkin and Kelly bags finished the year down 0.2 percent, which is to say flat, in a period when far more liquid assets moved by multiples of that.
Read together, these are not simply the old and expensive things holding up. Wine is old and expensive and it fell. What held value were the categories where the supply cannot be increased at will and where the object carries an account of who made it.
Richemont’s most recent quarter points the same direction from inside the industry. Group sales rose 11 percent at constant exchange rates in the three months to the end of December 2025, and its jewellery houses grew 14 percent, ahead of the group. Jewellery is the part of the business most often bought as a store of value rather than as a purchase, and it is outrunning everything around it.
Why this is happening now
Something changed underneath these numbers, and it did not start in the luxury industry.
Artificial intelligence has made a large category of output effectively free. Text, images, design drafts, music, code: things that recently took a team several weeks now take a competent operator an afternoon, and the cost of the tenth version is indistinguishable from the cost of the first. This is a genuine achievement, and it has a side effect that nobody in the luxury business planned for.
When one kind of production becomes infinite, whatever remains finite gets repriced. A hand-finished movement assembled under a loupe cannot be generated. Neither can a couture fitting across dozens of hours, nor forty years of a particular workshop’s accumulated habits. These things were always slow. What has changed is that slowness has become the distinguishing feature rather than an inconvenience the industry apologised for.
Scarcity used to be a question of how much of something existed. It is becoming a question of who made it, and whether a person had to be present at all.
It helps that this is now checkable. The growth of the resale market has had an underappreciated effect on the primary one: serial numbers, service records and auction histories mean provenance can be verified by a stranger rather than taken on trust. An industry that once relied on the buyer believing the story now operates in a market where the story can be looked up, which rewards the houses whose accounts survive inspection and quietly punishes the ones whose do not.
The part that could go wrong
Engineered scarcity is a choice, and choices can be reversed.
A house that has spent decades declining to expand can decide, under a new chief executive or a difficult quarter, that it will expand after all. Everyone holding the previous scarcity absorbs that decision immediately. This is a real risk and it is not hypothetical; the history of luxury is full of houses that discovered how quickly volume erodes the thing that made volume possible.
There is a subtler failure too. Scarcity works as a signal only while it looks like a consequence of standards rather than a marketing device. A waiting list that is obviously manufactured stops being flattering and starts being irritating, and customers who feel managed rather than served tend to leave with a story. Several houses have already found the line by crossing it.
The value in restraint, then, is not the restraint itself. It is the credibility of the reason behind it. That is a difficult asset to build, an easy one to spend, and impossible to generate on demand, which is precisely why the market has spent the past year paying more for it.
