iOS 18’s ‘big feature’ reportedly made with Google’s help, here’s everything we know

Apple used Google chips manufactured by Google’s cloud computing system for developing two AI models of its upcoming AI features and tools.

Apple relied on Google for developing two AI models for upcoming AI features. (Reuters)

Apple announced the launch of its suite of artificial intelligence features called Apple Intelligence at the WWDC 2024 conference this year. The features were expected to debut with the iOS 18 in the upcoming iPhone 16 series, however, beta updates suggest that Apple users will have to wait for the AI features. As a few of the Apple Intelligence features were made available to iPhone 15 Pro and iPhone 15 Pro Max users with the iOS 18.1 developer beta, a new report revealed how Apple's big rival Google helped the company to create these AI tools.

Apple trusting Google

According to a recent report by Reuters, Apple used the chips created by Google and not the ones created by market leader Nvidia for developing two essential elements of the artificial intelligence software infrastructure to be used in the upcoming set of AI features. 

Apple's move to trust Google cloud infrastructure is surprising as Nvidia is known for manufacturing high end and in demand AI processors. After including the other cloud computing companies who develop chips like Google and Amazon, Nvidia reportedly holds around 80% market share.

As per Reuters, Apple's recently published research paper did not completely deny the use of Nvidia chips. However, the details about the software and hardware infrastructure of building AI features did not find  mention of Nvidia anywhere. The company refused to make any comments following the release of its paper. 

The Cupertino based tech giant mentioned that it utilised two types of Google's tensor processing unit [TPU] which are assembled in large groups of chips. It used 2,048 of the TPUv5p chips for developing the AI model which will function on iPhones and other devices. 

Apple used 8,192 TPUv4 processors for making its server AI model. Nvidia is not a TPUs designer but the developer of graphics processing units or GPUs that are extensively used for AI tools. While Nvidia sells its chips and systems as independent products, Google sells TPU access via its Google cloud platform. The consumers who have an interest in purchasing access to TPUs are required to develop software using Google's cloud computing platform for using chips.

Reuters had reported Apple's dependence on TPUs in June but the exact extent of reliability on Google hardware was revealed after the coming out of Apple's research paper. Apple disclosed that it can make larger and more advanced models using Google's chips than the ones mentioned in the paper. 

Apple revealed a set of upcoming AI features during its developer conference in June. It also hinted at merging ChatGPT technology of OpenAI  with its software.