I'm thinking for example high end business line Lenovo, Dell, HP laptops where after 4-5 years they are out of any support and the failure rate is high enough that the best approach is to replace the whole thing with the newer model. Rolling out hardware takes time so you'll have 1-2 years of delay between when the new model hits the market and when it's in the hands of all users so eventually you just have a 4 year lifecycle with 25% of the hardware getting replaced every tear because the failure rate by then is anyway in the double digits (10+% of laptops in a company will fail one way or another in a year). You want to replace while it's still predictable and manageable, not when they cascade fail, because that also helps you stay modern, and not maintain a huge "just in case" stock.
You can't tell a user "wait while I quickly change your laptop's keyboard with this used one" because the manufacturer doesn't support the model anymore. It's not that much different with desktops although they split the failure rates across multiple different devices (separate screen, keyboard, the works). When you have (tens of) thousands of these you can't run around replacing memory modules, SSDs, screens, power plugs without it getting really expensive really fast. Not just the fix but the loss of productivity too. It's an experience I confirmed in every large company regardless what model or manufacturer they were using.
This leaves smaller shops, the kind that might buy no-name units, or even self assemble them, to look into "we can save by putting Chrome on these and maybe the later maintenance on that PC won't actually prove more expensive long term. Magically one PC will live for decades, I can confirm with mine :). But in bulk devices will fail at alarming rate after 4-5 years.
Yeah, a base model macbook air m2 is only $1100. You can't expect more than 8G of ram in that price range.