Agreed, this is crazy, and is exactly why the administration did what it did. However, every frontier model will require this within months.
It might be time for me to start looking into Chinese models or purchasing hardware for local llms, even if the cost amounts to 5-10k.
Im really not sure why anyone would spend 15k to run local llms. The models you'll be able to run (70b) param models will be incredibly underwhelming.
Is this assuming no price increase and no throttling ?