Apple May Use Google Cloud for Siri

This potential partnership would build on an existing, multi-billion dollar arrangement where Google pays Apple to be the default search engine in the Safari browser. That deal has been a cornerstone of their collaboration for years, despite the two companies being fierce competitors in the smartphone market. Apple has been quietly working on its own data center AI chip initiative, codenamed "Project ACDC" (Apple Chips in Data Center). This project, in development for several years, aims to leverage Apple's silicon expertise for its own server infrastructure, focusing on AI inference rather than training models. The company is also collaborating with Broadcom on another AI server chip, code-named "Baltra," expected by 2026. The move to consider Google's cloud for Siri comes as Apple's own AI infrastructure, known as Private Cloud Compute, has seen limited use, with reports suggesting only about 10% of its capacity is being utilized on average. Issues with the system, including slower update times and chips not optimized for large AI models, have reportedly hampered its performance. Historically, Apple has been hesitant to use Google's cloud for AI due to privacy concerns, with software chief Craig Federighi previously vetoing the option. However, changes to Google's security systems in 2023 reportedly satisfied Apple's privacy requirements, opening the door for the current explorations. This potential deal is part of a broader strategy for Apple to enhance Siri's capabilities by using Google's Gemini AI models. The arrangement is seen by some analysts as a way for Apple to quickly advance its AI features while it continues to develop its in-house capabilities. Apple's spending on data center infrastructure has been significantly lower than its tech rivals. In fiscal year 2025, Apple's AI-related capital expenditure was $12.72 billion, a fraction of the $50 billion to over $80 billion being spent by competitors like Microsoft, Google, Amazon, and Meta.