Why we hold no stock

Every distributor that holds stock is making a forecast: that in three to six months, customers will want this configuration, in this quantity, at a price that still covers what was paid for it. The forecast is made by people who are good at it. It is still a forecast.

What happens when the forecast is wrong

Wrong in the low direction is uncomfortable but survivable: the part is not there, the customer waits, or buys elsewhere. Wrong in the high direction is expensive. The stock sits, and server hardware does not sit well. A processor generation is superseded, a memory price falls, a drive capacity that was premium becomes standard. The book value falls faster than the shelf empties.

That loss does not vanish. It is recovered from margin on everything else in the warehouse, which means the customer buying the part that was forecast correctly is paying for the part that was not.

What happens when the forecast is right

Then the stock was bought at an earlier price, which in a falling market is genuinely cheaper, and in a rising or constrained market is genuinely valuable. This is the honest argument for inventory and we are not going to pretend it does not exist. It is strongest for parts with stable demand and a long life: rails, cables, optics, power supplies.

It is weakest for exactly the parts that make up the value of a server order — processors, memory, drives and accelerators — where a generation turns over roughly every eighteen months and where the configuration a customer wants is one of thousands.

The other cost of stock

The one nobody writes down: a supplier holding inventory has an interest in what it is holding. Ask a distributor with two hundred units of last generation whether last generation is the right answer for your workload, and you will get an answer, and it will be a good one, and you will never know how much the two hundred units contributed to it.

We have no units. When we say the older generation is the right answer for your workload — and quite often it is — the only thing that argument is serving is the argument.

What we do instead

We buy against the specification, per job, from whichever route delivers it: authorised distribution first, channel partners where a build-to-order or project price applies, and the professional secondary market where a factory lead time does not fit the project. Every line on the quotation says which route it came from.

The cost of this is speed on trivial orders. If you need one disk by tomorrow, buy it from a shop. The benefit shows up on the orders where it matters: a rack, a refresh, a build-out, where the difference between a specification that was thought about and one that was matched to available stock is worth considerably more than a delivery day.