Apple will be leaning on Google Gemini technologies Siri overhaul later in 2026, but it won't mean that iPhone user queries will be handed over and absorbed into Google's systems.

Apple and its AI strategy, Apple Intelligence, hasn't quite kept up with the rest of the industry, resulting in speculation that it would seek some outside assistance. On Monday, Apple finally struck a deal to do just that.

A multi-year deal between Apple and Google will see Apple use Google's Gemini models and cloud technology for Apple's foundational models. A statement received by CNBC, but not yet made public by Apple itself, reportedly confirms the deal has taken place.

"After careful examination, we determined that Google's technology provides the most capable foundation for Apple Foundation Models and we're excited about the innovative new experiences it will unlock for our users," wrote Apple in the statement.

The move has been suspected for some time, and could be an expensive relationship for Apple to maintain. In November, Apple was reportedly considering paying Google around $1 billion per year to use its models.

Apple has not commented on the deal's terms.

User data safety

To clarify, the deal doesn't mean that Google will be getting its hands on Apple's user data. It's one where Apple gets to use Google's Gemini models for its own processing, separate from Google's data-hungry systems.

The language of Apple's statement indicates that the Gemini models will be a foundation for Apple to build the Apple Foundation Models from.

Apple has been working on its own Apple Foundation Models for quite some time, with the intention of it being used to handle a new LLM-based Siri. However, as Apple fell behind the rest of the market, there was speculation that Google Gemini would be brought in to handle the queries.

Most people would assume that Google would be more directly involved in Siri queries as a result. However, Apple's history of protecting user data and privacy means it cannot do that at all.

Even with the current ChatGPT integration in Siri, the queries processed by OpenAI's system are not linked to the rest of a user's data. Privacy is ensured where possible, even here.

By using Google Gemini as the base for the Apple Foundation Model, Apple is able to take advantage of Google's existing work and build its own customized model.

In this way, Apple will keep user data stored on its own servers and take advantage of Private Cloud Compute for secure processing. Even if it's just Gemini in a Private Cloud Compute server, that data would be securely passed and discarded after use.

The Gemini models Apple will use won't be the same Gemini models you use on Google itself. Not by a long shot.

That customization may prove beneficial in other ways. For example, a deeper integration with App Intents in iOS. With greater access and control over apps, Apple could create a much more powerful Siri when it comes to performing tasks for users.

Apple's support for Model Context Protocol (MCP) will allow other external AI systems to make use of App Intents, too. But it won't necessarily provide as much capability as Apple's native system.