AI infrastructure is about to split into two geographies. Training clusters will migrate to the cheapest power on earth. Inference clusters will stay near cities. The companies that realize this first will gain a structural advantage in AI. I work in commercial solar. The energy architecture hyperscalers should be piloting now:
If VoS is below Austin Energy’s avoided marginal cost, then non-solar customers are not subsidizing solar customers. They are purchasing cheaper energy and grid benefits from them. Calling that a wealth transfer confuses who receives the bill credit with whether the utility is economically better off buying that output.
If a farmer purchases a game-changing piece of automated equipment that is going to dramatically improve his efficiency, but it requires assembly, is that farmer going to spend his entire life building that piece of machinery or will he spend a month or two mobilizing/deploying and studying it?