I dislike this kind of thinking in security -- it is a very 19th century.
The reason we only think about state security in the context of "geopolitics and military operations" is because that's all we hear about and that's probably all that nation-states used to be able to resource.
But with cloud + machine learning + sufficient sensors, the old assumptions about resource scarcity for security intelligence collection is flipped. It used to take a few "James Bond" agents, a few "Q" engineers, and a large support staff to work one case at a time. That scarcity calculus is completely different now. Instead of "1 civilian to every 300 police officers", new data collection feels more like "300 police officers to every 1 civilian".
The only thing that stops any sufficiently large organization from collecting dossiers on every person on Earth is "complexity versus competence".