> Mathematics was already not financially well rewarded,

Then why are frontier labs ploughing millions into racing human math researchers when they're already cash-strapped? What's the ROI for that?