A market economy is a distributed compute engine. The main reason why planned economies do worse historically is because the amount of compute needed to actually account for everything centrally is so immense (and thus costly), the implementers necessarily have to adopt some kind of simplified model, and then you get divergence between what the plan says and what's actually happening.

It's not a given that this will remain true forever, although I don't think it's tied to AGI. One could argue that AGI push is the trigger for a massive increase in compute capacity and corresponding decrease in price that might make this kind of thing viable, but that's just wishful thinking, not a fact.

Most of the definitions I've heard of AGI are that it outperforms humans at complex professional tasks. That would include capital allocation.