[D] Given the scaling up of deep learning methods, what are the remaining merits of staying in academia as an AI researcher?

Admittedly, I have worded the title question in a slightly naive and one-sided manner to instigate discussion. I see certain merits to academic labs pursuing deep learning research. However, it does seem that a lot of the big breakthroughs are now happening in industry labs, rather than in small university labs. This is likely due to DL maturing from an emerging research area into an industrial technology.

Given the recent developments in DL, what are people's thoughts on the relative …

academia academic big deep learning labs machinelearning research scaling scaling up