[FreezerburnV]

Because training costs are sky-high, and handling an individual request still uses a decent amount of energy even if it isn't as horrifying as training. Plus the amount of requests, and content in them, is going up with stuff like vibe coding.

If you want to know more about energy consumption, see this 2 part series that goes into tons of nitty-gritty details: https://blog.giovanh.com/blog/2024/08/18/is-ai-eating-all-th...

[sumeno]

The article says 80-90% of data center usage for AI is for inference, and is from a more reputable source than the random blog

[lkbm]

The blog is citing specific studies for its claims. Is there an issue with those studies?

[sumeno]

It's almost a year old at this point so at best it is horribly out of date