Are the switching costs of CUDA ecosystem potentially threatened because LLMs are now quite good at transcoding into other languages? In other words, is Nvidia'

Are the switching costs of CUDA ecosystem potentially threatened because LLMs are now quite good at transcoding into other languages? In other words, is Nvidia's greatest strength (AI) also potentially its undoing?