It's not really grayscale. The output of an image sensor integrated circuit is a series of voltages read one after the other, that could be from –0.3 to +18 volts for example, in an order specific to the sensor's red, green and blue "pixels" arrangement. The native picture (fig 1) is the result of converting a sensor's output voltage to a series of values from black (let's say -0.3 volts for example) up to white (let's say +18 volts for example) while ignoring if they are from a red, a green or a blue image sensor "pixel".

The various "raw" camera image formats kind of work like this, they include the voltages converted to some numerical range and what each "pixels" represents for a specific camera sensor setup.