8.2. Binary Floating-Point Promotion

Binary floating-point promotion is a special case of binary numeric promotion. It is applied to the operands of the exponentiation operator:

After the type conversion, if any, value set conversion is applied to each operand.