Rule of thumb
You have to develop a rule of thumb for kit, and stick to it. HARDWARE WILL DIE. This is inevitable. Don't waste your time trying to prevent it, simply plan around it. Have complete cold spare systems waiting in the wings, ready to be imaged and pressed into service.
For us, Kit can last two generations, and that's it. The first generation that kit is in service should be front-line use, high-demand applications where it matters. Once the warranty is up, if you have planned well you can often get another refresh cycle out of it by moving the gear to lighter duty. (Of course this is assuming you have appropriate spare systems and parts put away for the inevitable failures.) Our refresh cycles are three years, after the second refresh cycle, nothing is allowed to stay. Rework it, and give it away to staff, or a charity, if they're taking.
It's especially hard in a recession sticking to such rules; but you can't let the beancounters win. If you let the beancounters stretch out your refresh schedules, everything degrades. Your kit goes longer between refurbs, and the failure rate will climb. You probably won’t get a second cycle out of that kit, and you certainly don’t want to be pushing kit already on its second cycle any farther. Not only that, but the longer gear sits at front-of-line the less productive those staffs will be. Newer software is always more demanding, and just because you haven’t refreshed the hardware doesn’t mean they won’t be getting the latest version of photoshop, office, autocad or what have you.
Never, ever let the beancounters dictate IT purchase cycles, because the majority of them seem pathologically unable to understand soft costs. Too often they think about quarterly revenue instead of TCO issues such as support, maintenance, power consumption (and the associated cooling costs,) productivity loss due to slow gear, or incompatibilities. Every man-hour IT spends supporting a desktop is a man-hour not spend meeting some other business need, doing preventative maintenance, keeping our skills up to date, or doing research that may lead to operational efficiencies.
Moral of the story: don’t cheap out on the gear. If you plan it properly from start to finish, you can often get a second refresh cycle out of kit, but *only* if you have planned for it from the start, and *only* if that second refresh cycle sees the kit serving lighter duty. Deferring maintenance or refresh cycles can only end in tears.