I think there will be at least a minimum of focus on safety and value alignment.

They may have to focus on safety and aligning it to the companies values (else test/weak versions are likely to destroy/negatively impact the company or at least be useless). Which is at least something.

Only hyperrational/intelligent value unaligned intelligent systems would avoid this pressure. I don't see the development going that way.

It would still be sub optimal though.