Apple announced Monday a multi-year partnership with Google to power the next generation of Siri and Apple Intelligence features using Google's Gemini large language models and cloud infrastructure. The deal positions Gemini as the foundation for Apple Foundation Models after the company evaluated competing technologies from OpenAI, Anthropic, and Perplexity, marking a strategic shift for Apple's artificial intelligence strategy following repeated delays in delivering promised Siri capabilities.
The partnership involves Apple licensing Google's fully trained Gemini models while maintaining inference processing on Apple's own Private Cloud Compute systems and user devices rather than routing data through Google Cloud infrastructure. Apple stated after "careful evaluation" it determined "Google's technology provides the most capable foundation for Apple Foundation Models" and expressed excitement about "innovative new experiences it will unlock for our users." Bloomberg reported Apple will pay Google approximately one billion dollars annually for licensing and customized technical support.
The upgraded Siri powered by Gemini-trained Apple Foundation Models is scheduled to launch with iOS 26.4 in spring 2026, likely March or April. The new capabilities include understanding personal context by pulling information from Mail, Messages, and Calendar apps without explicit instruction, on-screen awareness allowing Siri to understand visual content the user is viewing, deeper per-app controls enabling complex multi-step actions across applications, and complex reasoning for queries requiring synthesis of multiple data sources like traffic, flight status, and wait times.
Apple's decision follows a troubled development history for AI-powered Siri features originally announced at the June 2024 Worldwide Developers Conference. The company initially planned to deliver enhanced Siri capabilities in iOS 18 but delayed the features multiple times. In March 2025, Apple acknowledged the personalized Siri features would not arrive until 2026, admitting development was "taking longer than we thought to deliver on these features." Software engineering chief Craig Federighi told employees in an August all-hands meeting that Apple had attempted to merge two separate systems—one for handling current commands and another based on large language models—and the hybrid approach failed to meet Apple's standards.
The partnership does not affect Apple's existing arrangement with OpenAI to integrate ChatGPT into Apple Intelligence for handling queries beyond Siri's capabilities. An Apple spokesperson told CNBC the company is not making any changes to the OpenAI agreement. However, industry observers expect Apple's reliance on ChatGPT will decrease as Gemini-trained Apple Foundation Models gain capabilities. ChatGPT will continue operating alongside Gemini for specific functions including image creation and editing through Siri and Image Playground features.
Google's involvement provides Apple access to Gemini 3, the latest iteration of Google's large language model series available in multiple versions with varying performance and price points. The most capable model, Gemini 3 Pro, costs four times more than the entry-level Flash edition. Apple stated it will use the large language model series alongside Google's "cloud technology," hinting the iPhone maker may leverage Google Cloud infrastructure including Tensor Processing Units for model training and optimization while keeping user data processing on Apple's own servers.
The deal represents a significant validation for Google's AI division and positions Gemini as the standard-setter for consumer AI applications. Alphabet's market capitalization briefly crossed four trillion dollars for the first time in intraday trading following the announcement, making it the fourth publicly traded company to reach the milestone after Nvidia, Microsoft, and Apple. Bank of America analysts stated the deal reinforced "Gemini's position as a leading large language model for mobile devices" and should strengthen investor confidence in Google's search distribution durability and long-term monetization strategies.
For Apple, the partnership provides access to advanced AI capabilities at attractive cost while theoretically maintaining privacy standards through Private Cloud Compute architecture. Wedbush analyst Dan Ives described the deal as providing Apple "a stepping stone to accelerate its AI strategy into 2026 and beyond." However, Apple's continuing need to rely on partners—first OpenAI and now Google—to deliver AI features raises questions about the company's ability to build competitive large language models internally despite massive research and development spending.
The integration will be white-labeled with no Google branding visible to end users. From a user perspective, the enhanced assistant will appear as Siri with dramatically improved capabilities rather than a separate Google product. Apple emphasized that "Apple Intelligence will continue to run on Apple devices and Private Cloud Compute, while maintaining Apple's industry-leading privacy standards," addressing potential concerns about Google's access to user data given the company's advertising-based business model and data collection practices.
The partnership is non-exclusive, allowing Apple to potentially integrate models from other providers including Anthropic's Claude in parallel. This multi-vendor strategy would preserve negotiating leverage and spread technological risks while requiring significant integration engineering to make different backend models seamlessly interchangeable. The flexibility to add additional AI providers suggests Apple views foundation model partnerships as infrastructure components rather than exclusive strategic relationships.
Industry analysts note the deal confirms that training state-of-the-art frontier models has become so capital-intensive that even Apple—the world's most valuable company with massive cash reserves—prefers licensing technology rather than building entirely in-house. CEO Tim Cook described the AI revolution in an internal meeting as "as big or bigger" than the internet, arguing for deep overhauls of research and development teams to catch up with competitors.
