I guess AI is not just LLM. Image processing, speech to text etc would fall under the use case. Regarding GenAI, Pixel phones already run nano model on the phone with decent performance and utility.