Let that sink in for a second... Apple just paid Google around a billion dollars a year to run Siri's brain.
The most valuable company on earth - the one whose entire mythology is
"we control the whole stack, the silicon, the software, the magic"
looked at the defining technology of the decade and decided it couldn't build a good enough version itself, So it rented one.
According to Bloomberg's Mark Gurman, Apple agreed to pay Google roughly $1 billion a year for a custom ~1.2-trillion-parameter Gemini model to power the rebuilt Siri's planning and summarizing. The partnership is now official, Apple and Google confirmed it jointly in January 2026.
In almost any other company, that's a humiliation you'd spend a year explaining to the board. For Apple, it barely moves the stock.
And that's the part worth sitting with.
“Why doesn't this hurt them?”
Everyone's asking
"Did Siri finally get smart?"
but that's the wrong question!
The right one is the one that actually decides who wins,the failure is real. Let's not pretend otherwise.
This wasn't a stumble. It was a multi-year, on-the-record, lawyer-involved failure:
June 2024 (WWDC): Apple previews a "more personalized Siri", one that understands your personal context, sees what's on your screen, and takes actions across your apps. It looked like the future.
September 2024: Apple runs iPhone 16 ads, including one with actor Bella Ramsey demonstrating Siri capabilities that had not shipped.
March 2025: Apple quietly admits the features are delayed. The phrasing - "we anticipate rolling them out in the coming year" - becomes a meme.
Tim Cook strips Siri away from AI chief John Giannandrea and hands it to Mike Rockwell, the executive who built the Vision Pro. A false-advertising class action gets filed the same week.
April 2025: The National Advertising Division (the BBB's self-regulatory body - not, as a lot of people online claimed, the FTC) tells Apple to drop its "Available Now" claims. Reporting at the time pegged the in-development Siri as working correctly only about two-thirds of the time. Not shippable.
June 2025 (WWDC): Still not ready. "In the coming year." Again.
2026: A reported ~$250 million settlement over the false-advertising claims, with eligible buyers getting up to $95 per device.
And this isn't a new-era problem. Siri has been the weak link since it launched in 2011. After Steve Jobs, the team reportedly settled into an annual update cadence, tied to iOS releases (a big mistake on current market IMO) while Google and Amazon iterated continuously.
The acquisitions that were supposed to fix it (who even remembers Topsy and VocalIQ?) got bolted on rather than integrated. Siri has been a punchline for over a decade.
So no, I'm not going to tell you Apple is secretly winning the assistant war. They lost it. Badly and publicly btw. For years, but as always, they find a way to do things the "Apple way"...
And to Apple, it doesn't matter!
Why did such a public failure not dent them?
“you might be asking yourself …’’
A bad assistant on one device is a reason to switch phones. A bad assistant inside an ecosystem you can't leave is just friction you absorb.
Look at the gravity well Apple sits in:
2.5 billion active devices, per Tim Cook on the January 2026 earnings call — up roughly 150 million in a single year.
~89% iPhone loyalty meaning when an iPhone owner buys their next phone, ~89% of the time it's another iPhone (CIRP, measuring the year to mid-2025).
Around 60% of Apple customers own three or four of the core devices — iPhone, iPad, Watch, Mac. Penetration runs roughly: iPhone 91%, iPad 75%, Watch ~66%, Mac 54%.
Services revenue north of $100 billion a year, at roughly 75% gross margin — about double the margin on the hardware itself.
Now picture the actual cost of leaving:
Your Apple Watch only pairs with an iPhone, so it's dead weight the day you switch.
Your iMessage threads degrade to green bubbles.
AirDrop
Handoff
iCloud photo library
Find My network that tracks your keys and your kid's backpack
All of it either breaks or gets worse the moment you step outside and not to mention the new parental-control features they just announced.
You don't switch an iPhone. You'd be switching a household.
That's why all this Siri fiasco can't pull anyone out. The lock-in compounds effect is real;
the model quality doesn't.
Siri being unhelpful, I won't say "dumb", is annoying. It's nowhere near annoying enough to make someone re-buy a Watch, re-train their family onto a new messaging app, and re-pay for a decade of apps. The moat was the ecosystem, not the model. And the ecosystem held while the model embarrassed them.
The part everyone's getting wrong: renting Gemini is the smart move (IMO)
This is where I disagree with the "Apple got humbled by Google" crowd. I think paying Google a billion dollars a year is one of the more clear-headed calls Apple has made in the AI era
The model is a deflating asset. Frontier quality is converging and inference prices tends to collapse across the board. Internally, Apple almost certainly evaluated OpenAI and Anthropic before picking Gemini, and it can renegotiate or swap providers later. When the thing you'd spend tens of billions to "win" gets better and cheaper and more interchangeable every single quarter, it isn't a moat. It's a commodity input.
Spending to win that race is overpaying for something that depreciates the moment you ship it. The best models of 2026 will be the cheap, boring middleware of 2028. (HOT TAKE).
So rent the commodity, own the appreciating asset. This is just good capital allocation. Pay ~$1B a year for a model you can replace (and use the traffic to train your own in-house), then pour the difference into the thing that compounds and that no amount of money can source: 2.5 billion devices, a twenty-year developer ecosystem, and the trust people put in the computer in their pocket.
It's not easy to build an ecosystem, and it's even harder to maintain one.
You can buy a model, You cannot buy a billion-device install base or two decades of switching costs. Failing at the model while owning the surface is survivable - Apple is living proof. Owning the best model while renting the surface - exactly the spot OpenAI and Anthropic are in - is the far more exposed bet. They have to earn every relationship from scratch. Apple already has them all.
Renting is a bridge, not a surrender. Apple didn't just plug Gemini into Siri; it built an architecture where the model is a swappable part. App Intents turns apps into things the operating system can call.
The Foundation Models framework (introduced at WWDC 2025, expanded in 2026) lets Apple own the interface developers code against - with a model-choice protocol behind it. Core AI lets developers run models locally on Apple silicon.
Put it together and Gemini isn't Siri's brain so much as a brain plugged into a socket Apple controls.
That's what changes how you should read the deal. Apple keeps training its own Foundation Models. The day they're good enough, Apple swaps them in behind the same socket and the ecosystem never notices which model is answering. Subsidize the model now, internalize it later, on your own timeline, while the surface keeps compounding the whole way.
The framework is the long game: own the interface, rent the implementation, bring it in-house when the economics flip.
So no, Apple didn't lose the AI race!
It looked at the race everyone's running, who builds the smartest model and decided that's not the one worth winning.
It's renting that part. What it's keeping is the thing nobody can buy.