Licensing requirements for AI & CUDA libraries during the transfer of equipment to the customer

Good morning,

I wonder about the license requirements, when the project is being transfered (sold as readdy to use equipmet) to client.

Let’s say that client requests for AI-based application that is procesing video source from few cameras in real time.
Let’s say that the solution includes:
Linux based pc machine with:
Nvidia RTX 2080 ti Graphics card
OpenCV with cuda Runtime
CUDA runtime
cuDNN Runtime
NVENCODE and NVDECODE runtimes
TENSORRT runtime
Custom AI based app that is dynamically linked with the runtimes.

Are there any special license requirements for the application or whole system should meet to be legally sold?