Confused about NoData values

What exactly is NoData? Since you can set NoData. Is it just a rendering after reading the data and finding areas where there literally is no data but there should be? For example a raster is constrained to 500×500, but the data only exists for 443×300, so during the read, it fills out the remaining area without data with whatever NoData is set as? Meaning QGIS would fill it out differently than ArcMap which would fill it out differently than GDAL since it happens during read? Or is NoData actually hard saved as those numbers in the file?

How can I access no data in GDAL? On the raster I’m working on, it seems to be in e notation 3.402823…e+38. If I wanted to do something like below with the band array to check for no data values, what is the appropriate thing to put in the if statement?

band = ds.GetRasterBand(1)
band_one = band.ReadAsArray()

for row in band_one:
    for column in row:
        if column is [*no_data*]:
            print("No data value found")