Gcc ignores cast to signed int16 and float ends up positive

ignored_cast.cpp

#include <iostream>
#include <cstdint>

int main() {
    uint8_t a = 0x80;
    uint8_t b = 0x01;
    float c = (float)(((int16_t)a << 8) | b);
    std::cout << c << std::endl;
    return 0;
}