Microsoft and Uber Are Running into an AI Cost Problem(firethering.com)
firethering.com
Microsoft and Uber Are Running into an AI Cost Problem
https://firethering.com/microsoft-uber-ai-coding-tools-more-expensive-than-human-workers/
6 コメント
I’m not convinced that Microsoft is terminating its Claude Code subscriptions from a cost perspective, but rather that they’d prefer teams dogfood their own GitHub Copilot tooling instead. This is similar to AWS pushing Kiro internally.
This is the actual answer. Very few teams internally actually got access to Claude Code to begin with. All teams still have access to the clause models. It’s a carve out to dogfood the agent.
Yea this makes sense but then why would they even bother to go with claude code earlier?
Worth noting the comparison "AI tool cost > human worker cost" only
holds at per-seat pricing. Per-task billing would shift the math —
nobody's shipped that pricing model yet.
"But but but "Jevons paradox" applies here as the cost of generating tokens falls, AI usage will dramatically increase meaning we need to build more data centers!"
Except that isn't true and the token cost isn't going down and even by building more data centers, these employees are still generating massive bills and Microsoft is upset that they have realized that its expensive.
The worst part is, this gives the illusion of "productivity" as employees waste more money and spend more tokens at the Anthropic casino.
Except that isn't true and the token cost isn't going down and even by building more data centers, these employees are still generating massive bills and Microsoft is upset that they have realized that its expensive.
The worst part is, this gives the illusion of "productivity" as employees waste more money and spend more tokens at the Anthropic casino.
I just don't understand why these tech billionaires wanna hype up everything so bad that even they know that people understand its just a unnecessary hype, yet they keep ignoring their core audience and keep on promoting more token usage or should i say wastage of natural resources. I use AI and I believe everyone here does at a certain extent its good but when something is forced, the conclusion comes out bad.