It’s both, a good state sets up the rules that help solve the game theoretic tragedy of the commons. Without a good framework and smart investment by the state, the 1000x more individuals in the private sector wouldn’t have been able to physically build the internet.

Government agencies like DARPA literally invented the internet, so it was definitely more than solving the tragedy of the commons, although that was important too.

DARPA is designed to limit how long they will fund any technology, and be very willing to terminate funding for any effort that isn't delivering. As I recall, even a 1% success rate is considered acceptable there.

The diffusion of information from innovation is a positive externality, so it makes sense there's a government role here.

But one of the pathologies you see in government-funded activities is sticking with funding because a group has become dependent on it, long after the effort ceased to make sense. Arguably nuclear is in that category now.