Nvidia is trying to increase its customer base. Look at the Mag 7, Amazon, Google, Microsoft, and even Meta are all working on their own inference chips. I don't think they can completely ditch Nvidia for LLM training, but they can make their own chips for inference. I believe that's also why OpenAI made its own.

No one wants to pay the Nvidia tax

If you make your own chip don't you need to make your own software stack too. Which is what I thought kept everyone using Nvidia.