Buying equipment too early is wasteful — a core economic constraint on capacity planning. Hardware steadily gets cheaper, faster, and more reliable (Moore's Law as the well-known instance), so waiting six months usually buys faster gear for less. Idle servers that drop in price before you use them earn you no goodwill from finance. This is just-in-time inventory applied to servers: Toyota cut cost and gained advantage by stocking parts only as needed rather than warehousing them. Installed-but-unused servers carry the same dead cost — rack space, power, setup time — plus Moore's-Law depreciation. The caveat is the startup exception: young companies rightly over-order out of fear, because an outage at launch costs more than the wasted hardware, and they lack the ops engineers to build a precise plan. As a company matures, the capacity process gets polished and JIT becomes the better discipline — buy later, not sooner, when your forecasts allow. --- *Source: [[The Art of Capacity Planning]] (John Allspaw, O'Reilly 2008) — Ch 4 — Predicting Trends*