I'm not sure how true that is for sectors other than software/AI, but in AI most academic researcher complain that all the moonshot research (often not even commercially applicable) is now solely coming from corporate research labs e.g. Deepmind, Facebook etc., as opposed to universities. They attribute this to poaching of faculty by companies, lack of resources in universities, lack of motivation in grad students to stay in academia.
"But in practice, BNNs do generalize to test points, and do seem to output reasonable uncertainty estimates. (Although it’s worth noting that simpler approaches, like ensembles, consistently outperform BNNs.)"