Use that to download the appropriate driver for your system.
You can download the latest driver that NVIDIA has available, for any of NVIDIA's GPU cards, except that at some point NVIDIA stops supporting a generation of systems, and then you cannot use a driver newer than the last one that supported your architecture. For example the newest drivers do not support Fermi architecture, if I recall correctly.
Note: if you happen to be using Mac, then there is no Tesla V100 driver available for any Mac OS, and probably never will be.
To do most GPU work, you do not need to download the CUDA development kit. However, you do need that kit to create custom GPU kernels, such as with the GPU Coder product. It usually does not hurt to install the kit, but you do not need it for simple gpuArray() operations for example. https://developer.nvidia.com/cuda-downloads