More seriously, it does seem like there were a number of interesting systems research and development collaborations in the 1980s: BSD at Berkeley, Athena at MIT, Andrew at CMU, etc.

Currently it seems like the interest, funding, opportunities, and incentives for academic researchers are largely for short-term projects and AI/ML rather than long-term, ongoing systems projects. The modern funding and publishing landscape seems to emphasize speed and quantity over quality and impact.

Moreover, it seems that companies with deep pockets (Microsoft, Apple, Nvidia) may be less likely to collaborate with and/or fund academic projects as IBM and DEC did in the 1980s. It could be that those partnerships weren't hugely beneficial for AT&T, IBM and DEC's businesses.