That’s straight from the standard. https://en.cppreference.com/c/numeric/math/hypot: “if one of the arguments is ±∞, hypot returns +∞ even if the other argument is NaN”

⇒ I expect this choice got careful consideration. (Edit: in the end, the reasoning still could have been “unfortunately, existing implementations do it this way; we’ll begrudgingly follow them”)