Anthropic doesn’t want you using Claude this much

I went pretty deep down a rabbit hole on Anthropic’s economics this week.

They’ve been one of the loudest companies warning about how powerful AI is becoming and pushing for more safeguards.

But there’s another incentive that I don’t think gets talked about enough:

Heavy AI users are expensive. Really expensive.

So I started looking at the overlap between Anthropic’s safety messaging, usage limits, and the actual economics of serving people who hammer Claude all day.

I ended up making a full breakdown because there was way more to it than I expected:

https://youtu.be/s1T9eaGt4lw

Curious what everyone here thinks... how much of the push toward tighter limits is safety, and how much is just the economics of running these models?

0 comments

Anthropic doesn’t want you using Claude this much