You can't really make that calculation with truly open ended, fundamental research. Both the potential benefits, and the complexity/cost of obtaining these benefits (or finding out they can't be obtained) can be enormous. At best, you can improve the odds of getting something rather than nothing, by e.g. hiring the smartest people and letting them do their thing with ample funding, a-la the Space Race or Manhattan Project.

It just pisses me off that we spend trillions on bombing random people in the desert, yet a few billion for e.g. fusion research is an insurmountable obstacle.

Well, any time you spend less than all of your money on fundamental research you're implicitly making a calculation of that kind.

Gambling is not "calculation".