If we can mine with pre-computed data, that is, the data is the mining machine, then even if the block reward decreases or the coin price decreases, it will not cause the hash rate to drop significantly. Because the cost of storing this data is very low.
I do not understand this or your proposal does not make sense.
Pre-computed data is just block hashes, mining is the process of scanning for a random block hash under a particular threshold. How do you take things that are already known, and turn them into something random?
Things that are random and have a sufficiently large range also naturally use more time to solve, because (again) the range is what makes mining so intensive, and expensive.
If you know Chia or Spacemesh maybe you can better understand my purpose. The same thing as Chia is that a part of the hash needs to be pre-calculated and stored on disk, and a VDF function is used. But the difference is that I think this part of the hash should not occupy too much disk space. Specifically, the cost of calculating this part of the hash should be much greater than the cost of storing this part of the hash. That is, use hash data as the mining machine instead of the disk. In this way, your mining machine neither needs to consume too much energy nor takes up too much physical space, because in fact they are just some calculated hash data. Of course it is very important not to let this data leak.