Every time a hot new camera is released, and the new Fujifilm GFX 100RF certainly qualifies, there’s a flurry of speculation online. Why is it in such short supply? Did Fujifilm intentionally limit production to create artificial scarcity? Are they playing games with the market?
I don’t think so.
What’s more likely is something far less sinister, but no less difficult: forecasting When a camera manufacturer introduces a product that breaks new ground—say, a medium format rangefinder-style body with high-resolution video and fast autofocus, there’s no historical model for how it will sell. There’s no previous data set to rely on, no safe analog to reference. That leaves product managers and operations teams trying to make tough decisions with imperfect information. Build too many units, and you risk unsold inventory gathering dust in warehouses, or worse, being discounted later and hurting your margins. Build too few, and you frustrate your best customers and miss out on potential sales.
There’s an old idea in operations research called the newsvendor model. It comes from a simpler world: a person trying to guess how many newspapers to buy in the morning, knowing they’ll be worthless tomorrow. But the core problem is the same: how do you make a decision under uncertainty when you only get one shot at it? The model formalizes the tradeoff between ordering too many and ordering too few. The more uncertain the demand, the more cautious the forecast tends to be. That’s especially true when the cost of overestimating demand is high.
For a camera like the GFX 100RF, the stakes are even higher. It’s an expensive, niche product with a novel form factor and an untested market segment. Even if internal enthusiasm is high, it’s risky to ramp up production without solid data on how it will actually sell once it’s in the wild. So companies hedge. They build a conservative number of units in the initial run, see how the market responds, and adjust from there. That might look like artificial scarcity, but it’s more often just caution in the face of uncertainty.
Here’s a Monte Carlo simulation of profit versus forecasted demand, which equals units built int he first run. Expected profit on each sold camera is $500. Expected loss on each unsold camera is $1500, which amounts to a pretty high writedown. Mean demand is 10000 units, with a standard deviation of 4000 units.
You can see that it’s easy to lose money if your forecast is too high.
Let’s make things more symmetric, with profit per sold camera and loss per unsold camera both $500.
In that case, within the range plotted, the camera maker isn’t going to lose money.
I have no idea what numbers the camera manufacturer was working with when they forecasted the GFX 100RF. But I do know this: forecasting is hard. Even when the tariff situation in one of your largest markets is, um, highly dynamic.
Leave a Reply