Swatch sold more in the first half and earned less. Net sales rose 8.5% at constant currency to CHF 3.12 billion, but operating profit fell to CHF 52 million from CHF 68 million — less than half what analysts expected. The distance between selling more and forecasting worse is the exact gap AI demand-planning promises to close.

The market read the currency story correctly — then sold the shares anyway. A stronger Swiss franc erased close to CHF 200 million of that momentum in translation, and no model prices the franc six months out. Currency explains the headline. It does not explain the rest.

The rest is a choice. Swatch kept its factories and its watchmakers running through the soft patch so it could answer a rebound without rehiring, which means it built this half for demand it had not yet booked. The operating margin slid to 1.7% from 2.2%, the arithmetic of a plant kept warm on faith. Idle capacity here is deliberate, and the cost of it came due.

This is the promise the vendors sell. McKinsey estimates that AI-driven forecasting can cut supply-chain errors by 20 to 50 percent, the number that turns excess inventory into a forecasting error you can fix. For a group carrying fixed costs in Swiss workshops against volumes that swing on mid-market mood, sharper sight sounds like the whole answer. The reflex is to file the shortfall under demand a better model would have caught.

Swatch did not misread demand; it overruled it.

Nick Hayek has said for years that he would rather book thinner profit than dismiss the people who make the watches. A forecast can tell him a rebound is late. It cannot tell him to let those hands go, and it could not have moved the number he was willing to accept.

Sales already reaccelerated in May and June, and Swatch says the pace held into July. The second half may simply confirm what the numbers are already suggesting. The forecast was never the part in doubt.