Apple is in talks to host its next-generation Siri chatbot on Google servers powered by Tensor Processing Units (TPUs), according to Bloomberg's Mark Gurman. The move marks a significant departure from Apple's emphasis on in-house silicon and its own Private Cloud Compute infrastructure.
Google's Gemini AI models will power Apple Intelligence features across the iPhone ecosystem, reaching approximately 1.5 billion users worldwide. Apple stated that after "careful consideration," it determined Google's AI technology "provides the most capable foundation for Apple Foundation Models." The deal signals a pragmatic shift: when workload is massive and quality matters, Apple is willing to rent best-in-class compute rather than wait for internal capacity.
The partnership validates Google's AI comeback after earlier struggles with Bard. For OpenAI, however, this represents a major loss—Apple was previously relying on ChatGPT integration. Analysts note this could hinder OpenAI's user growth and suggests "Google has edged past it in having the best AI models in the market."
