Apple officially announced that it has adopted Google's Gemini AI model to power the next-generation Siri.
Because two longtime rivals have joined hands in AI, a core technology, the decision is seen as more than a technology adoption: it is a signal that the market's structure is changing.
The market reacted immediately after the announcement.
Shares of Google's parent company Alphabet rose about 1.7% in a single day, pushing its market cap past $4 trillion, the second-highest valuation in the world after Nvidia.
Apple also drew a positive response, with the move seen as clearing up much of the uncertainty around its AI strategy.
- Why did Apple choose Google instead of its ‘own AI’?
Apple had been preparing an AI strategy built around its own foundation models, but its Siri upgrade schedule was delayed several times.
The AI-enhanced Siri originally targeted for 2025 was pushed back by about a year over technical maturity issues, while rivals' AI performance advanced rapidly in the meantime.
Apple's choice of Google comes down to three reasons.
First, performance that is ready to use right away.
Google Gemini is already a proven model in large-scale language processing and multimodal tasks, and can quickly raise Siri's ability to understand context and handle complex tasks.
Second, managing development risk.
Apple concluded that its own models alone could not meet user expectations in the short term, and it needed to buy time by using an outside model.
Third, privacy terms.
OpenAI, Anthropic and others were also considered, but Google reportedly accommodated Apple's privacy requirements most flexibly.
- The $1 billion deal: what the numbers mean
Under the deal, Apple is reported to pay Google about $1 billion a year.
This is less a simple licensing fee than a strategic cost for access to AI technology and long-term cooperation.
The technical gap is also large.
Apple's own cloud-based model currently in use is reported to have about 150 billion parameters, while Google Gemini can offer versions with up to 1.2 trillion parameters.
Even by simple math, that is a model size gap of nearly 8 times.
That is why people say Apple “borrowed outside technology to finish the job.”
- Siri's changing role with iOS 26.4
The Gemini-powered Siri is likely to be released with iOS 26.4 in March or April.
The heart of this update is not speech recognition accuracy but ‘understanding context.’
The new Siri is designed to carry out multi-step tasks by combining information shown on screen, earlier conversation context and the flow between apps.
For example, features the old Siri could handle only in a limited way, such as rescheduling, summarizing documents and passing commands between apps, are set to become fully available.
In this setup, Gemini serves as Siri's ‘brain,’ while Apple controls the interface and user experience.
- Using outside AI while protecting privacy
What Apple stressed most in this partnership is privacy.
Google's Gemini model runs only in Apple's Private Cloud Compute environment, not on Google's servers.
User data is not shared with Google and is not used to train the model.
It is a hybrid structure: Apple's own models still handle some basic functions, and Gemini is used only for advanced reasoning and summarization.
At the same time, Apple is developing its own model with about 1 trillion parameters and remains committed to reducing its reliance on outside partners over the long term.
- BITPRESS Insight: the signal this alliance sends to the market
This Apple–Google AI alliance is not about “who built the better AI.”
The key point is that even big tech companies have reached a stage where they struggle to handle AI on their own.
Going forward, AI competition
is likely to shift from a contest decided by model performance alone
to a race over how quickly and safely companies can absorb outside technology and blend it into their own ecosystems.
Another change worth watching is the basis for partnerships.
Where cost and performance used to come first, privacy and control over data have now become key factors in choosing technology.
Both individuals and businesses, when looking at AI services, have reached the point where
rather than just “which model is used,”
they also need to ask “who controls the data, and where is it processed?”
This alliance is a clear example that AI competition is moving into its next phase.
Sources
Compiled from CNBC, Bloomberg and Reuters reports
Official Alphabet and Apple announcements and market data
