A lot of mathematics research has no immediate applications but still is valuable. Software is more practical, governments should pay for creation and maintenance of "infra-structure software", which no single corporation is willing to pay for. Why aren't they? Because the end-results benefit everybody.

So I think the discussion about "better models to finance open-source" is misguided. Government(s) should pay for it, as much as it has a positive return on investment in terms of better infra-structure.

That is a matter of politics, not "better models". Will the government invest in infra-structure? That is a political question and largely depends on who you vote into power.