This is a most excellent place for technology news and articles.
My first use of Claude this week, for code reviews only(since no LLM can be trusted to write a user story or test suite), had it gaslight me.
It marked down my code for using a specific practice to make some xml safer and easier to read.
When I tried things its way, it wanted me to change it back.
I’ve used Claude and Codex, and while both are based on untenable economics, I can at least attest that my use of Codex has yielded some productive results. Claude, so far, has delivered fuck all that’s useful to me.
I have found the opposite. Codex spits back mostly useless code that is twice the length it needs to be with a bunch of unessesary stuff and Claude is the only thing I get useful output from.
looks inside
But if you use the $100 a month Claude Max plan, and you would use it to the weekly limit by going full ‘agentic coding’ (so almost no human in the loop) you would use an amount of tokens that would cost you more than $1000 at API-pricing.
If I watch 600 movies every day on my netflix subscription I am using more energy than I pay them for. Obviously everyone is like me. Therefore they are losing money overall.
Wait, their (netflix) earnings say they made a profit last quarter. But my calculations were waterproof!
Probably anthropic are not net positive, but they are not spending 10x what people pay them for tokens.
Except that you would need 50 devices to do that and the most expensive Netflix plan only lets you stream up to 4 devices at a time. Considering the average 2 hours per movie, that's 48 movies per day. That's without mentioning that you'd need to automate this because you'd be asleep for 8 of those 24 hours.
The point is, your analogy doesn't work. There's no reason why someone would do what you're describing and it'd also be very hard to do.
Using up all of your tokens though? Just use agentic coding, set the ""thinking"" to max and you'll see how quickly and easily you can burn through them. Share your account and you'll burn them even faster.
You're right that people can and do max out the expensive plans. Its very difficult to say how often. I just think a majority of anthropics customers are businesses, who often pay per token for easier scaling etc. According to the company, enterprise employees use about $150-$250 per month, (possibly max plans have similar use, which would support your view) but thats in API tokens which they probably have big margins on, so it's less likely anthropic are burning money on inference. If you want to convince me otherwise, its not enough to say that it can happen, it has to be frequent enough to outweigh the B2B sales. They are however likely losing money overall due to training costs etc.
I can't imagine paying for AI when the open source tools have made it so easy to set up a model locally.
