See, that’s not just what cloud is about.
If I want 2000 cores to say - crunch numbers real fast, for a day, once a month (e.g., I have some important business intelligence data I NEED to process ASAP once it comes in) - I can either buy a 2000 core cluster at millions of dollars (massive capex for stuff I won’t use 90% of the time), or I can rent it - and it becomes operational expense (that I can turn off and stop paying at any point) that I don’t need to raise a capex for and depreciate over 3, 5, or 10 years.
This is the advantage of cloud - I can scale compute capacity as and when required. Yes, its sometimes “cheaper” (and sometimes it seems so but you’re maybe comparing bananas to pears) to buy your own hardware, but you just can’t scale as quickly - lead time if I need to double compute in my cluster for example is months - chasing quotes, getting CAPEX approval, raising purchase orders, waiting for gear to arrive, etc.
Cloud? Adjust size of VM, reboot. Done. Sometimes not even that (e.g., trigger based apps; you pay only for the time your code runs, and it can scale as required).
You simply can’t do those things with your own gear.
And as to “monthly fees” - most businesses will be “charging” you a “monthly fee” internally for your division’s budget based on the depreciation of the gear you bought (i.e., accounts buy the thing, its meant to last 5 years so every month for 5 years, 1/60th of the purchase cost is taken out of your budget). You’ll be paying software licenses, storage costs, power, cooling, etc. You don’t escape “monthly” costs just because you run your own gear.
I’m not talking about home use here - I’m talking about enterprise.
Now, don’t get me wrong. Data sovereignty - I get it. I try to keep my personal data to myself as well. But if you’re bringing monthly fees into the argument as a reason to not use cloud… as above, it’s not that straightforward, especially for businesses where budgets, purchasing processes and tax/shareholder value implications are involved.