This is not the case at all; in fact Canon's actual sensor is on par with anything that Sony makes. One thing that you can easily see from the DxO data (if you know how to interpret it) is that the sensor itself is very quiet. The inherent noise of the sensor itself is as low as anything Sony or Nikon have ever produced. The trouble is the ADC architecture they have chosen to read the sensor data which appears to have a DR of around 12.5 bits (or stops).
Your stament is absolutely right, I did not use the right expresion. As you say the problem has to do with A/D Conversion and I guess it can not change without Canon going into patents issues with Sony.
Some people are also saying that canon has chosen to get good high iso instead of good base ISO, I do not think this is true.
The S/N ratio on 5D mkIII vs D800 is the same on all ISOs as is Tonal Range; DR is the same at high ISO but at base ISO is far superior on the D800, color sensivity is superior on the D800 at all ISOs and that is with a much higher pixel count.
The S/N ratio on 6D vs D600 is superior on the D600 up to ISO 12.800; DR is way better o the D600 up to ISO 400, about the same up to ISO 3200 and better on the 6D from that on, color sensivity is superior on the D600
6D is superior at ISO 25.600
And if you compare the 7d vs NEX-7 the 24mpx sony sensor is better at everything despite of the higher pixel count, specialy at DR and color sensivity.
So, in my opinion Canon sensors are way behind. OK, if you need best possible output at ISO 25.600 then get a 1DX or a D4, again the nikon will have best base ISO
If canon used a similar A/D conversion as Sony it would probably have the best sensors, or at least on par; but this will not hapen until they find a way of doing so without going into patent issues with Sony
All this started because it´s become normal for canon users to refer to DXOmark results as false; for camera reviews there are many different sites, but for sensor readings DXOmark is the only one I trust.