I agree on most of these points. It's like we're headed towards a fork in the road:

On the one hand, if AGI is expensive, then that empowers centralized bodies like governments and corporations. On the other, if AGI is inexpensive, then that empowers decentralized bodies, such as individuals and communities.

Plus, there's the question of agency and the diminishing returns of intelligence. If you have all the intelligence in the world but have limited ability to interact with the world, you only have so much agency. Nanotech enters the discussion here, but it's in such an early stage of development that we really have no idea what will be possible over the next decade or two, just like people in 2000-2010 had no idea what was coming in the 2020s for AI.