I don't think it will work like that. It is necessary to run the computation to get the result, even if you transfer the result back in time. So the world will end up running a lot of computations while knowing their future results. These computations will be a new kind of a tech debt, people will choose between adding one more task to the existing supercomputer (and slowing all the tasks there), or creating/leasing another supercomputer to finish some tasks early and forget of them finally.