Yes, but that’s resistance for potential users they can’t afford if they ever want adoption on the level of CUDA.