Ah, sorry, I'd misinterpreted the flow here. Yes, if this is randomly generated, then as long as there's enough entropy in the TPM-encrypted secret PBKDF2 is just an implementation detail - if there's ~2^128 bits of entropy in the secret to begin with, it's no easier to brute force that than it is to brute force the AES key to begin with.
Re: systemd-cryptenroll