Rumor
Apple Siri to utilize Google Cloud Platform and Broadcom-designed TPUs for Gemini integration
Monday, March 2, 2026 at 09:08 PM
Apple plans to utilize Google Cloud Platform to host Siri, which will integrate Gemini models running on Tensor Processing Units (TPUs) co-developed by Broadcom.
Context
Apple is reportedly in talks to migrate its next-generation Siri to Google data centers to handle an expected surge in AI demand. Despite significant capital expenditure, Apple’s proprietary Private Cloud Compute (PCC) currently operates at an average utilization rate of less than 10%, with much of its server hardware still uninstalled in warehouses. To support the spring 2026 launch of a more capable Gemini-powered assistant, Apple is leveraging Google’s hyperscale infrastructure to bridge its immediate internal capacity gap.
The partnership, estimated at $1 billion annually, represents a strategic shift in the AI supply chain. While Apple continues developing its own server chips and domestic data centers, the 1.2 trillion parameter demands of modern models necessitate third-party support for high-performance inference. For investors, this move signals a pragmatic pivot, prioritizing rapid market entry for a competitive AI assistant over total vertical integration. It secures Google as a primary infrastructure beneficiary in the high-stakes race for consumer-facing generative AI.
Related Companies
Apple
AAPL
Google
GOOGL