“You may not reverse engineer, decompile or disassemble any portion of the output generated using SDK elements for the purpose of translating such output artifacts to target a non-NVIDIA platform.”
The clause has sat in Nvidia’s online EULA since 2021. Since CUDA 11.6 it ships in the installed files too. The targets are ZLUDA and Chinese GPU makers like Moore Threads, whose MUSIFY tool moves CUDA code onto their chips. Nvidia’s moat is not the silicon. It is the license file.