Apple has taken a significant step toward deeper reliance on Google, reportedly asking the search giant to explore setting up servers that could run a future, Gemini-powered version of Siri. The move, first reported by The Information, signals a notable shift in Apple’s approach to artificial intelligence as it races to modernize its voice assistant.
The request comes just months after Apple confirmed in January that it would use Google’s Gemini AI models to power a more personalized Siri experience and other Apple Intelligence features. What was not made clear at the time was just how far that reliance on Google’s infrastructure might extend.
Apple’s Cloud Infrastructure Is Struggling
At the heart of the situation is a challenge Apple has quietly wrestled with for years: its cloud infrastructure has not kept pace with the demands of modern AI. According to the report, Apple’s internal AI setup is described as “beginning to decay.” The company was already in the middle of decommissioning older Nvidia-powered servers when it came to terms with how urgently it needed external support for cloud computing.
Apple currently runs its more complex AI tasks through a system called Private Cloud Compute, which operates on Apple silicon chips originally designed for consumer devices. Those chips are not optimized for large-scale AI workloads like running a model as powerful as Gemini. By contrast, Google operates its AI cloud systems on Tensor Processing Units — custom-built hardware specifically designed for running large language models. This hardware gap is one key reason Apple may be looking to Google’s servers.
What makes the situation more striking is that Apple’s Private Cloud Compute is barely being used. On average, only about 10 percent of its available capacity is in use, and some servers are reportedly still sitting in warehouses, never installed. Yet that underused infrastructure is still considered insufficient for the expected surge in demand once a Gemini-powered Siri goes live.
Privacy Was Once a Dealbreaker
Apple’s growing relationship with Google’s cloud is not without history. For years, Apple banned its AI engineers from using Google Cloud entirely, citing privacy concerns. Apple’s software chief Craig Federighi reportedly vetoed the option repeatedly. The tide turned in 2023, when Google updated its security systems in ways that addressed Apple’s requirements. After that, Apple quietly began adopting Google’s cloud for certain AI tasks.
The departure of key cloud talent, including Patrick Gates — the engineer who pioneered the idea of bringing Apple chips into data centers, which later became the foundation of Private Cloud Compute — also left gaps in Apple’s ability to build and scale its own cloud capabilities. The company has historically prioritized hardware devices and consumer-facing features over backend infrastructure, and that cultural reluctance has had consequences.
What a Google-Hosted Siri Could Look Like
According to reports, Apple has asked Google to investigate deploying servers that meet Apple’s privacy requirements — potentially inside Apple’s own data centers. This would allow Google’s AI systems to operate within an Apple-controlled environment rather than on Google’s broader public cloud network. The arrangement has not been finalized, and Apple is reportedly still evaluating multiple options.
One report suggests Apple is paying around $1 billion per year to Google to license the Gemini models for Siri. Running those models efficiently at scale would require the kind of hardware Google is best positioned to provide.
Apple has emphasized that on-device processing remains central to its privacy-first approach. Its Private Cloud Compute system also runs what it calls “stateless data processing,” meaning user data sent to servers for AI tasks is not retained after the response is returned — not even for logging or debugging. Any arrangement with Google would likely need to maintain those standards.
When Will the New Siri Arrive?
The upgraded, Gemini-powered Siri is expected to roll out in stages. An update arriving with iOS 26.4 will reportedly introduce more advanced features through Google’s foundational models. A more sweeping transformation — one that turns Siri into a ChatGPT-style conversational assistant — is expected with iOS 27, which is anticipated to launch in September alongside the iPhone 18 lineup.
The new Siri is expected to understand personal context by drawing on information from emails, messages, calendars, photos, and files. It may also respond to what’s visible on the screen and handle complex, multi-step tasks within apps — capabilities that would require significant cloud computing power to execute reliably.
For Apple, leaning on Google may simply be the fastest, most practical path to delivering a competitive AI assistant without bearing the full cost of building an entirely new cloud infrastructure from scratch.
