Apple Borrowed a (Chip) Helping Hand to Build Its AI

154

During a recent event, Apple CEO Tim Cook announced a significant partnership with OpenAI to integrate its advanced artificial intelligence model into Apple’s voice assistant, Siri. However, detailed technical documentation released after the event revealed that Alphabet’s Google has also played a crucial role in Apple’s AI developments.

To construct its foundational AI models, Apple utilized its own framework software in conjunction with a diverse range of hardware. This included Apple’s own on-premise graphics processing units (GPUs) and Google’s tensor processing units (TPUs), which are available exclusively through Google’s cloud services. Google has been developing TPUs for about a decade, and its fifth-generation chips, designed for AI training, offer competitive performance with Nvidia’s H100 AI chips. Google also announced that a sixth generation of these chips would be launched this year.

Google’s TPUs are specifically designed to run AI applications and train models. They are integrated into Google’s cloud computing hardware and software platform, enabling powerful AI processing capabilities. While Apple did not specify the extent of its reliance on Google’s chips compared to those from Nvidia or other AI hardware vendors, using Google’s TPUs typically involves purchasing access through Google’s cloud division. This is similar to how customers buy computing resources from Amazon Web Services (AWS) or Microsoft’s Azure.

Neither Apple nor Google provided immediate comments on this collaboration. The partnership highlights Apple’s approach to leveraging diverse and advanced hardware capabilities to strengthen its AI initiatives and enhance Siri’s functionality.