>>140234
Cloud design is the anti-thesis of efficiency. It's actually meant to trade off efficiency for liability. How can you even possibly conceive that there is efficiency in downloading 50GB of data to launch a single python program that orchestrates other python programs to run cold on 2 virtualization layers that load 500+ layers of overlay filesystems (even 10 was considered slow for user filesystems... lol) to create a coherent system to run a microscopic task that doesn't last a second but then needs its output piped to another such gargantuan abomination?

The point of all this crap is that if your userbase increases 10x, you just have to click a button and it will still work (except if you had used a monolithic single server you would have been able to host 100x the userbase before thinking of upgrading), and if your userbase dies off overnight, you can scale down without having invested in hardware that now sits idle with a big hole in your balance sheet (but not really because it would show up as depreciation instead in real life).

AI at least is definitely cool in practice (but none of the bullshit fluff that's been the only topic of attention since 2018) and has many amazing applications (if you don't fall for the 'chatgpt is sentient'-tier nonsense hype around "foundation models" and "LLMs" and all that crap).

But putting "cloud" and "tech" in the same sentence is worthy of a guffaw for sure. Ever heard the phrase "resume-oriented development"? Everything about cloud is that.