Apple May Use Google Cloud for AI-Siri Upgrade

This potential partnership is part of a broader, multi-pronged AI strategy for Apple, which is also testing Google's Gemini AI to power a more intelligent Siri. To accelerate its AI development, Apple has shifted from its traditional in-house approach to include partnerships with other major players, including a deal with Anthropic to use its Claude AI for coding assistance. The company has also integrated OpenAI's ChatGPT to handle complex user queries in its latest operating systems. This shift follows a leadership shake-up within Apple's AI division, with executives like Mike Rockwell, who led the Vision Pro headset development, now overseeing Siri's future. The move to partner with competitors like Google is seen as an admission that Apple has fallen behind in the AI race and needs to leverage outside technology to catch up. Apple is reportedly testing two versions of the new Siri concurrently: one powered by its own models and another running on external technology. The computational power required for training and running advanced AI models is immense, necessitating vast and expensive data center infrastructure. Training a 70-billion parameter model, for example, can require over a month of continuous processing on 32 high-end GPUs and can cost tens of thousands of dollars for a single training run. This has led to a massive infrastructure spending race among tech giants, with companies like Google, Microsoft, and Amazon investing hundreds of billions into data centers. Apple, in contrast, has adopted a more cautious approach to capital expenditure on data centers, spending significantly less than its rivals. The company's strategy appears to be a hybrid model, using its own Private Cloud Compute servers for some tasks while relying on third-party cloud providers for heavier lifting. This approach allows Apple to manage the anticipated increase in Siri usage without committing to the massive upfront investment in building out its own AI data centers.