Because the hard part isn't the compute, vector dbs, or whatever. It's the huge evergreen index of the whole internet. Getting over the hump of "every site lets your crawler work, gives good results to it, and bypass paywalls" is a massive barrier.