🌐 Summary: Cloud providers have to spend billions on hardware to run AI models for users. Apple will have no such cost because the model will run on… your iPhone.
🔍 In a recent publication Apple researcher show how they can run advanced AI systems similar to ChatGPT on computers with limited memory. Although not explicitly mentioned in the paper, it is clear that the iPhone is on their mind. There are several advantages for users:
🚀 speed – it always takes time to send a request from a phone to the cloud and retrieve the result. The first response will come faster when all calculations are done directly on the phone.
🔒 privacy – there is no privacy risk when your chat conversation with AppleGPT does not leave your phone. Apple has made privacy one of their main selling points. Already today the image recognition on your iPhone does not use the cloud. Now they can do the same for advanced AI systems.
💰 And then there is this other huge advantage for Apple. It is believed that most AI providers today are losing money on each request a user makes. Every image you generate and every question you ask adds to OpenAI’s debt. The cost of running AI models is insane, and the only one making money might be NVidia with their near monopoly on AI hardware.
💡 Traditionally software is a high margin business because the cost of adding extra users is very low. With AI this is no longer the case. Every additional user will require extra hardware investment to run the AI. Apple has 1.4 billion users worldwide. It would be impossible to give them all access to AI in the cloud. But Apple does not have to do that. They control the iPhone from which the request is sent and have now found a way to run the AI on the phone itself.
🤔 With AppleGPT, is Apple paving the way for a new era in AI accessibility and privacy? What are your thoughts?
