Huge result, but not P=NP
Many see refuting ETH as a major theoretical breakthrough, while stressing that subexponential is still far above polynomial and P vs NP remains open.
Doesn't transfer to all NP
Several point to the paper's note that 2^o(n) bounds don't survive general polynomial reductions, so it doesn't prove NP is subexponential or NP != EXP, correcting others who claimed so.
Little practical impact
Commenters note real-world SAT instances are already far easier than worst-case, so the algorithm is mostly theoretical; hopes for sub-quadratic attention are wishful.