I don't understand the CUDA lockin in 2026...

Surely anyone can take any piece of CUDA code and tell some LLM to port it to another platform and keep grinding till performance is identical?