This video was recorded before the rise of LLMs, making the specific language we use even less important. The reality is that the sunk costs of your existing codebase will almost always outweigh any advantage of trying a new language.
Sunk costs are an irrelevancy, relative future costs are what matters. Transition costs, especially for a mature product, may sometimes be an issue, especially if for some reason you are blocked from doing an incremental change.
And obviously neither irrelevant sunk costs nor relevant transition costs apply to a greenfield project.
Many (most?) projects will never qualify for such a rewrite, so JS/TS it is. I'm a huge fan of both, so I'm ok with them being around for a while.
In any case, the future is a judgment but the past is history --- and throwing more money at it won't change it or bring it back.