Google Cloud CEO Thomas Kurian revealed that Google will power core functionalities of Apple’s virtual assistant, Siri, utilizing its own servers. Speaking to “Handelsblatt”, Kurian stated that Google is building a model for “Apple Intelligence” which is an evolution stemming from Gemini, the name Google uses for its proprietary language models. Furthermore, Google will provide the entire infrastructure necessary for answering user queries directed at Siri from iPhones.
This deep collaboration with Apple, which had only hinted at a technological shift, is particularly extensive. Google will also be able to use Apple’s highly secure “Private Cloud Compute” service, specifically designed to process requests involving sensitive user data, all within Google’s own cloud architecture. Kurian was highly confident, noting that while Apple could have chosen any provider’s model, they ultimately selected Gemini. The expected rollout of the updated Siri has been postponed for over a year and is now anticipated for the summer.
Looking beyond the partnership with Apple, Google is also shifting its strategy for hardware sales. The company plans to directly sell its specialized AI chips, known as Tensor Processing Units (TPUs), to large enterprise clients, such as the hedge fund Citadel or the US Department of Energy. Previously, Google only rented these chips through its own cloud platform. Kurian confirmed that they are currently evaluating how to place TPUs in third-party data centers, calling it a potentially major business area. This push represents a more direct market challenge to chip industry leader Nvidia than ever before.
