Maybe this is not news to anyone, but I've repeated this statement enough times that I need this post to point at.
I share the point of view that LLMs are and will be something more akin to utility companies than a FAANG company. The models and running them represents infrastructure like an ISP or an electrical company. The LLM is an energy cost that we pay for in units called tokens.
The analogy I've been using is: Claude code is to Anthropic as cars are to gas companies.
We would never buy a sedan from Exxon, so should we be paying Anthropic to drive its own resource company?
How this is going so far
There are already cases where Anthropic have let Claude Code bugs run up usage to their financial gain:
I'd suspect this is only the beginning. Anecdotally, many of my friends and coworkers have used the past couple of weeks to upgrade their plans, setting them up for many months more of spend, whether Anthropic fixes this or not.
It's good that OSS options exist
Add more here 👆 if you'd like.
And that is all. This is simply to make the one point: Claude Code is a conflict of interest.
Update: As I continue to see examples of this surfacing, I've started collecting links which I'll leave here: