Apple Chose Google Over OpenAI: The $1B/Year Gemini Deal That Demoted Siri's Brain

Apple, which spent years insisting it could build its own AI, just agreed to pay Google $1 billion a year to run Siri, and OpenAI wasn't even in the running.

The Apple Siri Gemini partnership, announced in January 2026 and featured prominently at Google's Cloud Next conference on April 22, puts a custom 1.2 trillion parameter Gemini model at the center of Apple's rebuilt voice assistant. iOS 26.4 has already begun rolling out Phase 1 of the upgrade to compatible devices. The full conversational Siri, capable of handling multi-step tasks, maintaining context across weeks of interactions, and operating across every app in real time, arrives with iOS 27 and the iPhone 18 in September.

The deal's structure: Apple pays Google approximately $1 billion per year. Gemini handles the intelligence. Apple routes queries through its Private Cloud Compute framework, an architecture using end-to-end encryption and hardware-isolated enclaves, so that, according to Apple, no user data reaches Google's servers. The Gemini model weights run within Apple's own infrastructure.

What that arrangement reveals about who is winning the AI race, and how the decision was actually made, is more interesting than most of the coverage has suggested.

What Happened: Apple's $1 Billion Annual Bet on Google Gemini

Apple's AI problems are well documented. The company that built the first truly mass-market voice assistant with Siri in 2011 spent the following decade watching the product become a punchline. At WWDC 2023, Apple announced "Apple Intelligence", a suite of on-device AI capabilities meant to signal that it was serious about catching up. The announcement generated enthusiasm. The execution generated delays.

By late 2025, it was clear Apple's own large language models were not ready to power the kind of contextual, conversational Siri experience Tim Cook had implied was coming. Apple needed a partner.

OpenAI was asked and declined. According to reporting from 9to5Mac, OpenAI turned down the Siri integration because the two companies are now competing in the hardware market, OpenAI is developing its own AI device through a partnership with designer Jony Ive. Bringing Gemini to Siri while OpenAI is building a competing device created a conflict neither party wanted to navigate.

Anthropic was also reportedly considered and rejected for a different reason: price. Anthropic reportedly sought "several billion dollars a year" to power Siri, a number Apple found untenable for a transition product that the company still intends to eventually replace with its own models.

Google offered performance, a privacy architecture that Apple could credibly market to its users, and a price Apple could accept. The custom Gemini model it developed for Apple operates at 1.2 trillion parameters, significantly larger than most publicly available frontier models. The capability improvements are measurable: the Gemini-powered Siri reportedly achieves a 92% completion rate on complex multi-step tasks, up from 58% on the prior version, with response times under half a second.

The rollout proceeds in two phases. Phase 1, now underway with iOS 26.4, enables screen awareness, Siri can see and interpret content in any app in real time, along with a 128,000-token context window, expandable to 1 million tokens, that allows it to remember conversations across weeks. Phase 2 arrives with iOS 27 in September, bringing the full conversational Siri 2.0 experience Apple has been building toward.

Why This Deal Reshapes the Foundation Model Market

Google won something larger than a vendor contract. Both Apple and Samsung now use Google Gemini as the AI backbone of their respective voice assistants. Those two companies together account for more than 80% of the world's smartphones. Gemini is not just a product people can choose to use, it is the AI layer that runs on the devices most people already carry.

The strategic effect on OpenAI is significant. Before this deal, ChatGPT existed as an optional integration in iOS, users could invoke it deliberately for tasks Siri couldn't handle. With the Gemini upgrade, that opt-in role shrinks further. Gemini handles the core Siri experience; ChatGPT handles whatever Gemini cannot. For the vast majority of users who never change defaults, this means their daily AI interactions are now routed through Google, not OpenAI.

Distribution at scale matters more than model quality in consumer markets. The best model in the world, accessed through a deliberate app download, reaches a fraction of the users reached by the second-best model baked into the operating system. OpenAI built its growth on ChatGPT's viral adoption. The Apple deal makes that path narrower.

For Apple, the arrangement is an admission the company has been careful not to make explicitly. Apple Intelligence was marketed as a native intelligence system. The headline feature of that system, Siri, now runs on a competitor's model. Apple's framing is that this is a strategic transition while its own 1 trillion parameter model, expected to be ready in 2027, is developed. That model is a real project. Whether it will be capable enough to replace Gemini on schedule is a real question.

The deal also validates a market dynamic that AI labs have been racing to capture: OEM integration. The path to scale for foundation models is not convincing individual users to download your app. It is convincing device manufacturers to build your model into the operating system. Samsung chose Gemini for Galaxy AI. Apple chose Gemini for Siri. Both are multi-year commitments. Google now sits inside more than 80% of the world's smartphones, and the agreements run for years.

The Part Apple Isn't Advertising

Apple's entire brand position on privacy rests on a simple proposition: your data stays with you, not with advertisers. Google's entire business model rests on a different proposition: data about what people do is the raw material for the world's largest advertising operation. These two propositions are difficult to make compatible.

Apple's answer is the Private Cloud Compute architecture. Queries to Gemini are processed through Apple's own infrastructure using end-to-end encryption and hardware-isolated computation. The Gemini model weights, the mathematical parameters of the model itself, run on Apple's servers, not Google's. According to Apple, no user data is transmitted to Google. The experience of using Siri 2.0 on an iPhone, in Apple's account, does not give Google access to what you said or what was on your screen.

That claim is technically specific and plausibly true. It is also, for some observers, difficult to evaluate from the outside. Privacy architecture is only as trustworthy as the implementation, and users have no independent means of auditing whether the described safeguards are functioning as stated. For users managing sensitive research or work notes who prefer their data to stay on-device, remio's local-first knowledge base offers a privacy-conscious alternative to cloud-routed AI tools.

Elon Musk made this point publicly. When the Apple-Google deal was announced, Musk criticized it on X, raising concerns about the concentration of AI power in a single provider. His specific concern: Apple and Samsung both using Google creates a situation where a single company's AI models are processing interactions from more than 80% of smartphones globally. That degree of concentration, Musk argued, creates risks that apply regardless of any individual privacy claim. It echoes concerns regulators raised a decade ago about Google's dominance in search, a domain in which Google also claimed its practices were neutral and beneficial.

The competitive dynamics that led to this deal are also revealing. OpenAI declining the contract is not a footnote, it is evidence that the relationship between Apple and OpenAI is more complicated than it appears. OpenAI, through its partnership with Jony Ive's design company io, is developing a dedicated AI hardware device. Apple, through its AI hardware roadmap, is building the infrastructure that would make that device unnecessary for its own users. These companies are no longer partners building complementary products. They are, in meaningful ways, competitors.

Anthropic's rejection is different. Anthropic reportedly priced itself out of the deal. That pricing reflects a real capability premium, Claude is widely considered the enterprise market's strongest model, but it also reflects a B2B strategy that prioritizes professional and enterprise users over mass-market consumer devices. Anthropic has almost no consumer distribution. It didn't get Apple, and it didn't get Samsung. Whether that gap matters to Anthropic's long-term position depends on whether enterprise market share remains more valuable than OEM integration as AI matures.

Samsung, Google, and the Race to Control the AI Layer on Every Phone

The historical parallel that fits most cleanly here is not recent. In the early 2010s, Google paid Apple approximately $1 to $3 billion per year, a figure that eventually grew to an estimated $15 billion annually, to be the default search engine in Safari. That arrangement made Google a verb in the lives of iPhone users while generating enormous revenue for Apple. It was a deal in which Apple's distribution and Google's technology produced mutual benefit, and regulators eventually asked hard questions about whether it constituted an anticompetitive arrangement.

The new deal runs the same basic structure in reverse. Now Apple is paying Google for the AI technology, and Google is paying for distribution through the relationship. The scale is smaller, $1 billion per year versus $15 billion, but the structural logic is similar: Apple controls the device, Google controls the intelligence, and together they reach the world.

For Samsung, the decision to use Gemini for Galaxy AI reflects similar reasoning. Samsung's own AI research, while serious, is not at the frontier of large language model development. Rather than invest decades and tens of billions to build a foundation model, Samsung is licensing one. The result is that every major Android OEM and Apple now share the same AI brain.

For OpenAI, the device market is the strategic response. If OEM integration is the winning distribution path, building a device is how you create your own OEM. The io partnership with Jony Ive is the company's answer to a distribution problem it didn't fully anticipate when it launched ChatGPT as a web product in 2022.

iOS 27, Apple's Own Model, and the $1 Billion Question

The Gemini-powered Siri upgrade is not Apple's final position. The company has been explicit that it is developing its own 1 trillion parameter model, targeted for completion sometime in 2027. If that model performs as Apple hopes, it becomes the basis for renegotiating or exiting the Google arrangement. Apple has strong reasons to want its AI capability in-house: margin, user data (which stays in Apple's ecosystem rather than flowing through an arrangement with a competitor), and the brand coherence of being able to say Siri is genuinely Apple's.

Whether the 2027 model arrives on schedule, and whether it matches Gemini's performance, are open questions. Apple's recent track record on AI timelines, Apple Intelligence was announced in 2023 and delivered in partial, delayed waves, gives analysts reason for caution.

In the near term, the September 2026 launch of iOS 27 and iPhone 18 is the first real test. The full Siri 2.0 experience, conversational, contextual, proactively helpful across apps, will be what most users experience as the first genuinely useful version of Siri. If it works as described, it will be the most significant product change Apple has shipped since the App Store.

The broader competitive stakes will clarify over the next 18 months. If Apple's own model is ready and capable in 2027, the Google deal becomes a transitional chapter. If Apple's model is delayed or underperforms, the $1 billion annual payment becomes the baseline for an ongoing dependence, and Google's leverage in any future negotiation grows accordingly.

The apple siri gemini deal is, at one level, a straightforward technology licensing arrangement. Apple needed a better AI engine; Google had one; they agreed on terms. At another level, it is a map of where AI power is consolidating. One company's models now run inside more than 80% of the world's smartphones, with multi-year commitments from the two largest device ecosystems on earth.

That's not a competitive outcome. That's a structural fact about how AI infrastructure is being built, and who gets to sit at the center of it.