The accounting "magic" is that they can claim the savings today, but that's just normal amortization.

And this will increase their server failure rate, but probably not enough to move the needle in terms of the random failure buffers they account for.

Data centres fill up. They fill up with important "production" software and associated data housed in the same DC for locality, which never moves and mustn't be shut down. After a while the data centre is full and everyone needs to put their new software in the newly built empty data centre.

Over time a data centre gets technology upgrades later or not at all (say, a new-new networking fabric is invented and starts rolling out before the new networking fabric finishes rolling out). There's a slow decay in terms of service. New data centres have on-site offices of engineers who keep everything running at maximum efficiency, older data centres have a single security guard who receives a daily printout of which machines to press the reset button on while doing their rounds. Eventually an old data centre is turned off all at once.

If they're keeping older servers around longer, they may be delaying turning off the least valuable computers, continuing planning upgrades to 'new' even after 'new-new' is ready and rolling out, or slowing down the invention of new tech. As I understand it, "decommission X many machines due to age" isn't really the model.

I had the impression that servers were vastly more expensive than network equipment. I wonder what the breakdown is for network equipment.

Not writing down investments is how Enron committed it's mark-to-market fraud.

A $1000 server with a 4 year life span is expensed at $250 a year.

A $1000 server with a 6 year life span is expensed at $167 a year.

You convert your server lifespan from the former to the latter you get $83, or 33%, savings a year.

If you (intend to) keep your servers for 6 years and do not make this adjustment, then you're doing a false accounting, not the other way around.

When I was in banking, table stakes involved being able to reconstruct a company's financials from the footnotes. The data to double check the output are all there. That lets you undo these assumptions, if you choose to, and replace them with your own. (99% of the time, the effect is de minimus.)

I was stuck supporting one of these environments, and made a business case to overbuy the hardware a few years ago when they did a “free” socket to core migration. They’re still running those things, as the maintenance on the servers is less than paying for more cores.