Apple Intelligence is out in Beta now, and it has become public knowledge that Apple didn’t use any Nvidia AI hardware to train its AI models. Instead, Apple used Google’s TPU chips.
Apple published a new research paper outlining how it developed and uses its AI models, and the omission of any mention of Nvidia hardware is interesting, to say the least. Nvidia’s stock has been growing insanely fast on the back of its AI push, and many thought that it would be the lone supplier of AI-related hardware for years to come.
The research paper says that Apple used 20,48 Google TPUv5 chips to build its AI models and 8,192 TPUv4 chips to run its AI servers. Back in 2023, Apple shared that it plans to invest $5 billion in its own AI server enhancements, so it will remain to be seen if it sticks with Google’s hardware or moves to something more proprietary in a few years. Since this is Apple’s first big AI push, I wouldn’t rule out Nvidia. The American tech giant might still build its own data centres, which could use tech from a variety of suppliers.
If you want to see where Apple places its models against others like Chat GPT and Meta’s Llama-3-8b, you can check out the full Apple research paper.
Source: Apple Via: Max Weinbach
MobileSyrup may earn a commission from purchases made via our links, which helps fund the journalism we provide free on our website. These links do not influence our editorial content. Support us here.
