So are TensorFlow and PyTorch. All AI/ML frameworks have to translate high-level tensor programs into executable artifacts for the given hardware and they're all given away for free because there is no way to make money with them. It's all open source and free. So the big tech companies subsidize the compilers because they want hardware to be the moat. It's why the running joke is that I need $80B to build AGI. The software is cheap/free, the hardware costs money.