tesla just became the latest company to cut back on ai spending imposing a $200 per week token spending limit for employees and pushing them to use grok + cursor models - sign of the times to come

> elon's pushing employees to use internal versions of grok + cursor model composer.
> they've spun up multiple private agents trained on tesla data to boost productivity
> employees still get access to claude and gpt via their internal "bottleneck platform".
> this follows uber, microsoft and meta who've now cut back aggressively on spending
note that elon probably operates the largest compute cluster colossus I,II & III so this implies a plot of employees were leaning heavily on competitor models.