by studying the algos that create those wallets and patterns that may exist i think that number will be much lower, add to that if somehow you found thousands of people who are willing to brute force using a good algo for couple of years straight, then what are the chances that you maybe find a wallet with a large balance,
The complexity of those would still be fairly high, unless a weak RNG is used which could possibly lower the keyspace due to it's non-random generation. That's what some of the bruteforcing projects are attempting to do by getting these weak keys.
and for people who may say this theory is crazy and not possible i refere you to the guy who actually done just that for couple of years and found private keys of 3 small wallets.
Possibly weak keys.
This is like digging for gold and by technology going up i believe oneday it will be possible to brute force about 20-100 wallets each year using some crazy asic machine and a whole industry will be built around just this idea.
this is just a theory of mine which made me curious, would love to hear what you think other than no it's impossible.
Current ASIC cannot be used to generate addresses but it should be fairly simple to do so as well. Other than the speed (which we have established that it would have to generate addresses extremely quickly), you'll have to compare the addresses generated to those addresses that currently have any Bitcoins which could present itself as a slight bottleneck depending on the way it gets designed.