Apple quietly rewired the AI map this week: the iPhone maker confirmed a multi‑year deal to use Google’s Gemini models as the brain for a major Siri upgrade coming in 2026. The move is simple on paper and seismic in practice — it hands Google a high‑value licensing customer and hands Apple (and its users) a fast track to more capable generative AI, while promising to keep user data on Apple’s own infrastructure.
The headlines and the dollar signs
Apple and Google described the collaboration as multi‑year and emphasized that Gemini will run on Apple devices and on Apple’s Private Cloud Compute to protect privacy. Independent reporting has filled in the financial sketch: multiple outlets have suggested the arrangement could cost Apple in the ballpark of $1 billion a year, with the Financial Times calling it a multi‑billion dollar cloud computing contract over time.
That price tag, if accurate, makes this one of the largest commercial LLM relationships to surface so far — and a big revenue and cloud win for Google. It also amplifies Gemini’s reach: powering Siri would put Google’s models on hundreds of millions of iPhones in addition to Android devices.
Why OpenAI didn’t get the job
In a twist, the Financial Times reports that OpenAI made a conscious decision not to pursue or accept the role of Apple’s custom model provider. Sources say OpenAI opted last autumn to focus inward — on its own device ambitions and building next‑generation models — rather than become Apple’s bespoke supplier. That matters because Apple had already been working with OpenAI in other ways, integrating ChatGPT for particularly complicated Siri queries and Apple Intelligence features.
So this wasn’t simply Google outbidding a rival; it was a strategic choice by OpenAI and a commercial lift for Google. Analysts quoted in coverage called it a major setback for OpenAI and a landmark win for Google’s AI momentum.
What this means for users and the industry
For iPhone owners the promise is straightforward: a more conversational, capable Siri that can generate ChatGPT‑style answers and handle more complex tasks. Apple has delayed its big Siri upgrade before — saying last year the rollout would take longer — but this deal signals the company is now committed to shipping those improvements in 2026.
Apple’s insistence that the Gemini models will operate on its private cloud compute (and partly on‑device) is meant to blunt privacy concerns. Still, the arrangement concentrates a lot of AI horsepower behind two corporate walled gardens: Apple’s hardware and Google’s models — a configuration that will attract scrutiny from competitors and regulators alike. Remember, Google already pays Apple billions to be the default search engine on iPhones, and courts have recently allowed such commercial arrangements to continue.
Gemini’s rise helps explain Google’s appeal. The model family has been woven into Google’s consumer and workplace products, and its growing presence across services is part of a broader strategy to lock in usage and cloud revenue. That momentum is also visible in efforts like Gemini’s deeper search integrations across Gmail and Drive, which show how Google is knitting LLMs into everyday productivity tools Gemini’s Deep Research may soon search your Gmail and Drive.
Apple’s choice also reflects a pragmatic posture: the company has avoided becoming a pure AI arms‑dealer, preferring to assemble best‑in‑class partners while keeping control over user data and device experience — a role one FT piece described as Apple playing kingmaker between the cloud AI giants.
If you want a granular look at Apple’s move to use a custom Gemini model for Siri, that earlier coverage lays out the technical and timing cues that led here Apple to Use a Custom Google Gemini Model to Power Next‑Gen Siri. And observers should watch how Google packages agentic features and cloud services for partners — those capabilities helped make Google the attractive supplier it is today Google AI Mode adds agentic booking features.
The immediate effect: better Siri for billions of users and a lucrative new revenue stream for Google. The longer game is more interesting — we’re seeing the architecture of consumer AI being written in commercial contracts, with winners determined as much by partnerships as by model benchmarks. Apple kept privacy and control on the table; Google bought scale and distribution; OpenAI chose a different gamble. The AI chessboard has new pieces — and the game just got harder to predict.
By the way, if you’re thinking about Apple hardware while this unfolds, the company’s ecosystem — from iPhone to Mac — is the context where much of this will land. For those shopping around, the latest MacBook models remain a solid way to stay in that world MacBook.