Apple adopts Google’s Gemini to power next-generation Siri and AI

Apple is partnering with Google to rebuild the core of its artificial intelligence offerings, announcing a multi-year agreement to base its forthcoming Apple Foundation Models on Google’s Gemini large language models and Google Cloud infrastructure. The companies said Apple selected Gemini after a "careful evaluation" and concluded the technology "provides the most capable foundation" for the next phase of Apple Intelligence, including a more personalized Siri expected later in 2026.

Under the collaboration, Gemini models and Google Cloud will serve as the foundational layer for Apple’s internal models, which in turn will power a hybrid architecture of on-device processing and Apple’s Private Cloud Compute. Google emphasized that the arrangement will leave Apple Intelligence running on Apple devices and on Apple’s private cloud and said the partnership will maintain Apple’s "industry-leading privacy standards." Apple framed the move as a way to accelerate features it first previewed at WWDC 2024 but put on hold after delays announced in March 2025.

The technical pivot marks a significant course correction for Apple, which had been developing its own generative models and integrating third-party systems for selective tasks. The new Gemini-backed models are expected to substantially upgrade Siri’s generative capabilities, turning the assistant into a more proactive, conversational answer engine and reshaping how millions of iPhone and iPad users search for information and interact with their devices.

Investors greeted the announcement as clarification of Apple’s AI strategy. Wedbush analyst Dan Ives called the deal "an incremental positive to both," reflecting analyst views that the partnership reduces uncertainty about Apple’s path to large-scale AI deployment and boosts Google’s cloud business by adding a marquee client with massive reach.

Many commercial details remain undisclosed. Apple and Google did not reveal financial terms or the degree of customization Apple will demand. Industry discussion has centered on previously reported figures suggesting Apple considered paying roughly $1 billion per year for a custom Gemini variant running on Apple’s private cloud, but that number has not been confirmed by either company. Nor did the announcement specify which Gemini model variants will be used, the degree to which Apple will modify models for its ecosystem, or the technical partitioning of tasks between device and cloud.

The agreement also raises open questions about Apple’s current use of other generative engines. Apple has integrated OpenAI’s ChatGPT into parts of Siri and Apple Intelligence for certain complex queries; the statement did not clarify how those integrations will coexist with or be replaced by Gemini-based systems.

Apple said the Gemini partnership will help power future Apple Intelligence features that are slated to roll out later in 2026. Some product timing expectations point to an appearance of the next-generation Siri with iOS 26.4, possibly in March or April, though Apple has not provided a firm release date.

For now, the announcement signals a pragmatic turn: Apple will lean on a leading external model to accelerate features it has promised users while keeping critical processing and privacy controls within its own devices and private cloud. The real test will come as engineers and regulators scrutinize the technical details of data flows, model customization, latency and privacy protections once the companies disclose how Gemini will be integrated behind the familiar face of Siri.