The spares decision that matters is not which parts to stock. It is which parts will still be buyable in year eight — and what you do about the rest.
Most spares lists are built from failure probability: what breaks, how often, how long it takes to get. That is necessary and insufficient, because it treats availability as constant. Over a decade it is not — components go end-of-life on a schedule that has nothing to do with your machine.
A durable strategy splits the bill of materials three ways. Commodity items that will be buyable indefinitely need no action beyond a normal stock policy. Items with a known obsolescence horizon need a plan: a qualified alternate, a redesign, or a last-time buy sized to the remaining life. Custom items — your fixtures, your panels, your assemblies — need a documentation guarantee more than a stock guarantee.
That last category is where a manufacturing partner matters. If the build record, the released revision and the test procedure still exist, a replacement can be built in year eight that matches the one installed in year one. If they do not, the "spare" is a reverse-engineering project with a machine standing still.
The practical step is unspectacular: review the bill of materials against obsolescence risk once a year, and act on the few lines that changed. It takes an afternoon and it removes the class of surprise that costs the most.
— GANI Lifecycle engineering team