I like seeing proposals like this that use Bitcoin as essentially a protocol layer (and in this case value store).

40KB anything is not going anywhere near the blockchain soon; this is going to be a no-go for the dev team and miners.

There are also a bunch of ancillary questions, like can these zero knowledge proofs (presumably non-interactive ones) be combined up with the rest of the blockchain to be turing-complete? Also a no-go.

Anyway, this is cool. Given current Bitcoin decision making processes, I would expect it would need a solid year of great adoption in some sort of side-car process before it had a shot at main blockchain integration, and even then, it would have to get drilled down to 1 or 2k of data max.