He was making a sweeping statement that anything triggered via eBPF can be proprietary, which is surely not correct. But, the "game changer" that he was talking about was getting proprietary code running inside the kernel using this method..