>LLMs are expensive and are only going to become more expensive

Prices per token are dropping roughly 10x per year for the same quality of model. Bills are going up because AI agents use way more tokens per task, not because the models cost more. The losses come from training new models and buying hardware, not from selling tokens too cheap (Anthropic's inference margins are estimated somewhere in the 60–80% range).

>As you’ve probably heard a million times by now, the counterargument to not hiring juniors is the question of succession. Who will be your future seniors?

As if companies cared that much for the long term?

If they did they wouldn't have made 90% of their stupid ass product and market "quick money grab / alienate users" decisions.

The answer to the question of "future seniors would be the same as with those decisions: "who cares, us C-levels would have moved on by then anyway" and "we'll handle it when we get there".

[dead]