Apple announced at its Worldwide Developers Conference that the upcoming Siri AI will be powered by Google’s Gemini language models, hosted on Nvidia hardware inside Google’s data centers. The company emphasized that, even though the models run on Google’s cloud, Apple’s privacy safeguards remain unchanged and Google receives no access to user data.
The move marks a shift from Apple’s traditional approach of keeping AI processing on‑device or on its own private‑cloud servers. Apple has long highlighted on‑device analysis and encrypted cloud services to keep personal information out of reach of both Apple staff and third parties. However, the size and capability of modern language models exceed what can be run locally on iPhones or Macs, and building a dedicated Apple‑run data center would be a massive undertaking.
Apple asserts that all data transmitted to the Gemini service is encrypted end‑to‑end, preserving the same privacy guarantees it has offered for years. Analysts will be watching how the partnership holds up to those claims, and whether future updates expand Siri AI’s abilities while maintaining the same level of user confidentiality.
