#seL4 is based on amazing ideas, but I have yet to see an ecosystem built around it.

I think there are a few reasons:

  • It requires an MMU, so a lot of low-end microcontrollers are excluded.
  • Many of its users are in the defense sector. They are likely either unable to contribute to open source or have huge overheads for doing so.
  • While the kernel is GPLv2, modifying it is very strongly discouraged, and contributing changes is very difficult because you have either not break or fix the proofs.
  • Some of the most exciting potential uses are blocked by the poor support for speculative execution mitigations.