For big metal rigs like supercomputers, there is almost no choice but Linux. Combined with Linux' dominance of the server market in general, there's a risk emerging to get stuck in a monoculture on the OS layer.

Most companies, including very big ones with a long history of OS development are abandoning their efforts in favor of just using Linux. The perception is often that there is no innovation that can be done on the OS layer that will earn back its investment.

This leads to a lack of experimentation, tunnel vision, and in the long run stagnation.