To be fair, launching their own ChatGPT wouldn't be that much more interesting than leveraging OpenAI. What would be amazing is if they get a high-quality GPT model running locally on the device and integrate that with their services.
Unfortunately very few Macs would be able to run anything like that since 8GB is still the base config for consumer models and I wouldn't be too surprised if 80-90% of all Macs sold in the last couple of years only have 16 GB or less.
Now.
As you correctly point out, their margins on RAM are quite high.