Yes, right. We could completely erase one third of exploitable vulnerabilities (by your numbers) only by switching to modern languages.

There is no good argument against that. Why wait for C or C++ to try and implement get another weird “solution” for those problems? (That no one uses then anyway)