I cross-posted this in the Linguistics comm (!linguistics@mander.xyz), with a few additional comments. Might as well share them here:
There's an earlier study from the Max Plank Society that estimates the family to be 8k years old (this one gives makes it 6kyo). That study was using a glottochronology technique; it gives you a decent guess about the age of a family/branch, but this one gives you actual data instead.
And yet both kind of mix the Caucasian and Steppes hypotheses - with Early PIE in the Caucasus, Late PIE in the Pontic-Caspian Steppe.