For years, Siri has been the AI assistant that Apple users tolerated rather than loved. You could ask it to set a timer. You could ask it about the weather. And if you asked it anything more complicated than that, you’d usually end up typing the question into Google yourself.

Apple knows this. And instead of trying to build a smarter Siri from scratch, they did something nobody expected: they called Google.

What Apple Just Announced

iOS 26.4, expected to roll out this month, will include a fundamentally rebuilt version of Siri powered by Google’s Gemini AI model. This isn’t a minor upgrade or a new voice pack. It’s a full architectural overhaul, the kind of change Apple almost never makes to a core product.

The deal reportedly cost Apple around $1 billion, and the result is a Siri that can actually think.

Here’s what that looks like in practice:

On-screen awareness. The new Siri can see what’s on your screen and act on it. If you’re reading an email about a flight, you can say “add this to my calendar” and Siri will pull the details from the email, create the event, and set a departure reminder. No copying. No switching apps. No “I’m sorry, I can’t do that.”

Multi-step task chains. You can now give Siri a complex request like “Find the cheapest flight to New York next Friday, add it to my calendar, and text Sarah my arrival time,” and it will execute the whole chain. Up to 10 sequential actions from a single command. That’s not an assistant, that’s a workflow engine.

Real conversations. Siri can now maintain context across a conversation. Ask a follow-up question and it remembers what you were talking about. This sounds basic, and it is. It’s also something the old Siri couldn’t reliably do.

The Uncomfortable Truth

Let’s be honest about what this deal actually means: Apple, one of the most valuable companies on Earth, couldn’t build a competitive AI assistant on its own.

This is the company that designed its own chips, built its own operating systems, created its own cloud infrastructure, and controls every pixel of its hardware. But when it came to making Siri smart, it had to go to its biggest competitor for help.

Apple originally announced these Siri improvements back in June 2024, targeting iOS 18. Then it slipped to iOS 26. Then to iOS 26.4. Bloomberg reported in February that some features are being pushed even further, to iOS 26.5 (May) and iOS 27 (September). Voice-based in-app controls and the ability to search through personal data, like finding old messages or podcasts friends shared, are reportedly still not ready.

The fact that Apple had to delay features it announced two years ago, and then still needed Google’s help to deliver them, tells you everything about the current state of the AI assistant race.

Why Google Said Yes

This might be the more interesting question. Google already has its own AI assistant. Why help Apple build a better one?

Money, obviously. A billion dollars is a billion dollars. But there’s a strategic angle too. Every time an iPhone user asks Siri a question and Gemini processes it, Google learns something. Every query, every follow-up, every task chain. Apple says all of this runs through its privacy framework, with on-device processing where possible and encrypted cloud connections when not. But the model itself, the intelligence, lives in Google’s hands.

Apple is essentially renting Google’s brain. And Google is happy to be the landlord, because the rent is paid in both cash and data flow.

The Privacy Question

Apple is leaning hard on its privacy reputation here, and for good reason. The company knows that its users chose iPhones partly because Apple promised to protect their data. Handing the keys to Google’s AI could undermine that entire brand promise.

Their answer is a hybrid architecture. Most basic tasks stay on your device. When something requires Gemini’s full reasoning power, the request goes to Apple’s Private Cloud Compute infrastructure, which then connects to Gemini through what Apple describes as an encrypted, anonymized pipeline.

Will this actually protect user privacy? Maybe. Apple’s track record on privacy is genuinely better than Google’s. But “we send your data to Google through an encrypted pipe” is a harder sell than “your data never leaves your device.” And the details of what exactly gets shared with Gemini, and what Google retains, haven’t been fully disclosed.

What This Means for You

If you use an iPhone, here’s the practical reality:

Short term (March-April 2026): iOS 26.4 will bring the first wave of Gemini-powered Siri features. On-screen awareness and multi-step tasks should be available on devices with A17 Pro chips or newer. Older iPhones will get limited functionality.

Medium term (May-September 2026): More features will roll out with iOS 26.5 and iOS 27, including deeper personal data integration and in-app voice control. These are the features Apple keeps promising and keeps delaying.

Longer term: This is probably the beginning of a permanent partnership. Apple doesn’t have the AI research bench to compete with Google, Anthropic, or OpenAI on its own. Expect Gemini to become a deeper and more permanent part of iOS with each release.

The Bigger Picture

Apple giving Siri to Google is a signal about where the AI industry is heading. Building a world-class AI model isn’t something you can do with money alone. It requires years of research, massive datasets, and a specific kind of engineering culture that Apple, despite all its resources, doesn’t have.

The companies that built strong AI foundations early, Google, Anthropic, OpenAI, Meta, are becoming the infrastructure layer for everyone else. Apple, for all its power, is becoming a customer.

That’s not a failure. It’s a recognition that the AI landscape has consolidated faster than anyone expected. The question for the rest of us is simpler: will Siri finally be worth talking to?

Based on what we’ve seen, the answer is probably yes. But it won’t be Apple you’re really talking to.