Given that this is something Zuckerberg said 12 years ago, I'd give him the benefit of the doubt and hope that he doesn't think that way anymore. (The benefits of experience!)
Not relevant to the original discussion or your point, but I've heard Fortran is still in use and pretty performant (and readable) for what it's being used.
Yes, between legacy software (powering expensive and irreplaceable legacy hardware), and a userbase that doesn't care so much about learning the fancy newer tools as long as they can do their experiments, I think it's still standard in many parts of physics.
