Long term reliability metrics can only be gathered for technologies that have already been deployed in the long term. That’s why many mission critical applications like this one use “obsolete” technologies.
...or perhaps it's a much more simple risk probability v. consequence severity decision?