That’s a fair question. I can’t claim to have anything that would merit being called a “metric”. The number of papers and projects using Julia increases every year, but that in itself doesn’t quite support my claim. What I mean is that there are regularly new large-scale computational projects adopting Julia, and they are the types of projects that, in most cases, I’m pretty sure would have used Fortran seven years ago.

I guess this project list is a possible metric.

https://juliacomputing.com/case-studies/

The fact that Fortran applications are not collected in a single page by a for-profit company that does not exist in the case of Fortran, does not in anyway prove that new applications moving away from Fortran.