Apple Intelligence used to rely on Google

The developer beta with Apple Intelligence (well, a waitlist for it) is rolling out, getting many people hyped. And now, a new artificial intelligence research published by Apple reveals that the Cupertino company has used Google hardware for the early foundations of Apple's generative AI project, Apple Intelligence.

The research paper is called "Apple Intelligence Foundation Language Models" and details the known sources of the language model that is at the base of Apple's generative AI. What's interesting is a quote buried inside the paper that hints that Apple may have been using Google's hardware in the early development stages.

The paper says Apple's Foundation Model (AFM) and the server tech that it uses were initially built on "v4 and v5p Cloud TPU clusters" using Apple software. A CNBC report suggests that Apple rented time on existing Google-hosted clusters of servers; however, the research does not directly say that or mention anything about Google or Nvidia. It's more likely that Apple bought the hardware from the company and used it in its own data centers.

However, using Google hardware in the early stages of making Apple Intelligence doesn't influence too much the future of the tech. Apple is reportedly planning to optimize AI applications within its own data centers and is said to have its own hardware based on Apple Silicon for its data centers to process Apple Intelligence queries.

Reportedly, Apple is planning to allocate over $5 billion to AI server enhancements over the next two years. Cupertino is aiming to match the tech capabilities of Microsoft and Meta in AI and reportedly plans to acquire tens of thousands of AI server units.

From all of this, I can see Apple is really taking generative AI seriously. Also, the iOS 18 Apple Intelligence features that the company introduced during its development conference are making me seriously excited about Apple's take on generative AI. It seems, at least to me, that Apple is in a great position to quickly overtake generative AI leaders in the market, at least if all the features it releases work as intended.