The real reason ePBS will have limited impact, is because it aims to fix trust issues from the wrong side. The difficult trust problems aren't with the proposer trusting the relays or block builders. We already see that the vast majority (90%) of proposers trust out-of-protocol sidecars to receive bids
The real trust problem that ePBS can't address, is the trust that order flow originators need to have of builders. Originators need to trust that those to whom they are sending flow, will not exploit that flow, for example by sandwiching it, unbundling transaction bundles, replacing backruns with their own, leaking private flow accidentally, etc. On top of which, each new builder adds outgoing bandwidth costs and complexity that, for a new builder landing few blocks might not be worth it to the originator. This is why you cant just spin up a builder the same way you can a proposer and build profitable blocks.
If we want to have a decentralized block building space, that's the side of the equation that needs to be solved.