Apple Surrenders AI Core to Google: Gemini Will Power the New Siri
It’s official: Apple has chosen Google Gemini to power the next generation of Siri. Here is why OpenAI lost the deal of the decade.
Apple and Google have entered a historic multi-year partnership to use Google’s Gemini models as the foundation for the next generation of Apple Intelligence. After years of trying to build a rival AI infrastructure, Apple has effectively handed the keys to its most critical software product, Siri to its fiercest competitor.
This isn't just a chatbot integration. Unlike the optional ChatGPT toggle added in iOS 18.2, this deal makes Google’s technology the core engine driving Apple’s upcoming "Personalized Siri" and its broader Foundation Models.
The Deal: A "Complete Foundation"
In a joint statement released Monday, the companies confirmed that Apple’s next-generation Foundation Models will be built on Google’s Gemini architecture and cloud infrastructure.
While the exact terms remain confidential, analysts from Deepwater Asset Management and reports from the Financial Times estimate the deal will see Apple pay Google between $1 billion and $5 billion annually to lease its AI infrastructure.
This flips the script on their long-standing search arrangement, where Google pays Apple approximately $20 billion a year to remain the default search engine on Safari. This time, the money flows from Cupertino to Mountain View to secure the iPhone's future.
"After careful evaluation, Apple determined that Google’s AI technology provides the most capable foundation," the companies stated.
Why Google Won (And OpenAI Lost)
For the last year, OpenAI’s ChatGPT seemed poised to be Apple's primary partner. This deal ends that speculation.
While ChatGPT remains available as a third-party extension, Google has won the infrastructure war. The deciding factor wasn't just model intelligence, it was scale.
- TPU Dominance: Google owns the entire stack, from custom TPU chips to data centers. Apple needs to serve 1.5 billion daily Siri requests. OpenAI simply lacks the compute capacity to handle that volume at the speed and cost Apple demands.
- The Admission: This is a rare admission of defeat. Apple’s internal "Ajax" model and "Private Cloud Compute" initiatives seemingly failed to reach the quality needed for a true Siri overhaul by the 2026 deadline.
What This Means for Your iPhone
The "New Siri," powered by Gemini, is expected to launch later this year (likely in the iOS 20 cycle). Users can expect:
- True Context Awareness: Siri will finally understand what is on your screen and take action across apps.
- Multimodal Capabilities: The ability to process video, audio, and text simultaneously in real-time.
- Speed: Running on Google's infrastructure likely solves the latency issues that plagued early Apple Intelligence demos.
The Privacy Firewall
Apple is acutely aware of the optics: Google is now powering the iPhone's brain.
To mitigate privacy fears, Apple confirmed that while Gemini provides the logic, the execution still happens inside Apple’s "Private Cloud Compute" (PCC) environment. User data will not be used to train Google’s models, and personal requests are purportedly anonymized before processing.
Whether users will trust Google with their personal context is the multi-billion dollar question. But for now, the verdict is in: The new Siri is Google.
