Cuda application deployment on Windows

I developed a C++ Windows desktop application using Cuda, and each time I want to use it on a new computer, just installing the latest NVidia drivers don’t work, I also have to install the Cuda SDK. If I only install drivers, then all the results from the Cuda kernels are 0s without any error message.

Is this normal to have to install the Cuda Developer Kit everywhere even if I just want to use the application that has been compiled? Otherwise what is the minimum things to deploy with the application in order to have it working on other computers than the one it has been developed with?