Organoids are a perfect example of Michael Levin's "intent all the way down", of cells self-organizing. During development, stem cells either get allocated into different types of developed cells based on chemical (and probably 'electrical') gradients - or they figure it out on their own, recruiting each other into types, tissues and roles. The first is fairly standard developmental dogma (good book on how that applies to the CNS here); Levin's lab has been showing that it's more the latter.

To get a flavor of the complexity of the brain, I’d very much recommend Mark Humphries The Spike (he’s here on Medium, IMO worth a follow).

Agreed that however they’re organized, nerve cells have managed to be much more energy efficient than attempts to emulate them in silica / software. If you buy one of the observations in Jeff Hawkins’ Thousand Brains that cortical stacks are somehow passively predictive, each stack for multiple ‘frames’ of prediction, and fire (and so use energy) only when a prediction is unmet somehow (disclaimer: I do) then we have the beginnings of an explanation of the multiple orders of magnitude difference in energy efficiency. I’m not a neurologist, but the obvious candidate for this kind of passive prediction is the interconnection between pyramidal cells (large number of both dendrites and axon terminals).

That said, organoids are highly variable in makeup and organization from one to the next. Great overview for anyone who might want to start in this area here.

It’s a fascinating area. but beware — it’s a deep rabbit hole to fall into. :-)