AI
The Apple Google AI Deal Is the Most Lopsided in Tech History
Apple is paying Google approximately $1 billion per year to license a custom Gemini model that will power Siri on 2 billion devices. The math, the structure, and the language of the deal say something everyone is missing: Apple lost.
The Deal Apple Hopes You Don't Read Closely
The Apple Google AI deal is the most important structural story in consumer technology this decade. The Apple Google AI deal commits Apple to paying Google approximately $1 billion per year for a custom 1.2 trillion parameter Gemini model that will power Siri on roughly 2 billion active Apple devices.
The deal was announced in January 2026. Google Cloud CEO Thomas Kurian confirmed at Google Cloud Next '26 in Las Vegas: "We're collaborating with Apple as their preferred cloud provider to develop the next generation of Apple Foundation Models based on Gemini technology. These models will now power future Apple Intelligence features, including a more personalised Siri coming later this year."
Every word of Kurian's statement is doing work. "Preferred cloud provider" positions Google as the senior partner. "Apple Foundation Models based on Gemini technology" makes clear whose technology is underneath the Apple-branded wrapper. "These models will now power future Apple Intelligence features" demotes Apple's own foundation model efforts to past tense.
The press treated this as a partnership. It is not a partnership. It is Apple outsourcing the brain of its most important product line to its largest competitor — and paying for the privilege.
The Per-Device Math Nobody Ran
Apple has roughly 2 billion active devices in the installed base. If Apple is paying Google approximately $1 billion per year for Gemini access, the per-device cost works out to:
$1,000,000,000 ÷ 2,000,000,000 devices = $0.50 per device per year
That number is striking in two directions.
In one direction, it shows Google is giving Apple a price that no consumer or enterprise customer could ever access. OpenAI's ChatGPT Plus consumer subscription runs $20 per month, or $240 per year. Anthropic's Claude Pro is $20 per month. Apple is getting frontier-model access for 0.2% of consumer pricing per device per year. That's the kind of pricing that requires Google to view this as a strategic loss-leader rather than a profitable commercial deal.
In the other direction, $0.50 per device is comically cheap relative to the strategic stakes. If Gemini-powered Siri performs well, Apple keeps 2 billion devices in its ecosystem and prevents the Android exodus that an inferior AI assistant would trigger. If Gemini-powered Siri performs badly, Apple's hardware advantage collapses. The downside is measured in tens of billions of dollars in lost device revenue. Apple is paying $1 billion to mitigate a $50 billion risk.
For Google, the math is similarly asymmetric. Gemini's training and inference costs are now subsidized by Apple at a thin per-device rate, but more importantly, Google gets distribution access to 2 billion devices it could never reach through Android alone. The deal is a strategic checkmate against OpenAI, who had been positioned as Siri's AI brain through a 2024 announcement.
Why the Apple Google AI Deal Demotes OpenAI
In June 2024, Apple announced that ChatGPT would be integrated into Siri as part of Apple Intelligence. The announcement was the biggest endorsement OpenAI had ever received from a consumer hardware company. Sam Altman was on stage. The framing was that Apple had picked OpenAI as its AI partner of choice.
Eighteen months later, that framing is gone.
"OpenAI's ChatGPT integration in Siri remains active but is clearly the backup, not the primary brain."
The structural implication for OpenAI is severe. The company is heading into its own IPO in September 2026 at an approximately $852 billion to $1 trillion valuation. A meaningful portion of that valuation depends on OpenAI being the default consumer AI brand — the company everyone thinks of first when they think of AI. Losing the primary Siri integration to Google is the kind of event that should affect the IPO multiple.
The market hasn't priced this in because the Apple-Google deal is being read as additive ("Siri will use both Gemini and ChatGPT") rather than competitive ("Gemini is replacing ChatGPT as the primary Siri brain"). The reality is the latter. The phasing makes this explicit.
Phase 1 (Spring 2026) — Gemini provides context awareness in iOS 26.4.
Phase 2 (September 2026, alongside iPhone 18) — Full Conversational Siri runs on Gemini, with multi-turn dialogue and complex task completion.
ChatGPT is not mentioned in either phase. It's present in the OS as a legacy integration that users can invoke explicitly, but the default Siri voice — the one that responds when you say "Hey Siri" — is Gemini-powered from Phase 2 onward.
The Apple Google AI Deal Is Apple Admitting It Lost
Apple has built its entire identity on vertical integration. The chip, the OS, the apps, the services, the retail — all designed and controlled in-house. The company explicitly rejected the model of relying on third parties for core technologies. Tim Cook has spent a decade telling shareholders that Apple's advantage comes from owning every layer of the stack.
The Apple Google AI deal is the first major exception to that philosophy in fifteen years.
Apple Intelligence, announced at WWDC 2024, was supposed to be Apple's answer to ChatGPT and Gemini. The company invested billions in building its own foundation models, in custom silicon (the Neural Engine in M-series and A-series chips), and in privacy-preserving on-device inference. The pitch was that Apple's vertical integration would let it build AI that was both private and powerful.
That pitch failed. Apple's own foundation models could not match Gemini or GPT-class performance. The on-device privacy story was real but solved a problem most users didn't prioritize. The company that built its brand on doing everything in-house had to license its most important new product feature from its largest competitor.
The financial impact of this admission is real but not catastrophic. Apple's services revenue continues to grow. iPhone sales remain strong. The hardware business doesn't depend on Apple winning the AI race on its own. But the strategic message is significant: in the next era of consumer technology, Apple is a hardware company that licenses software intelligence from Google. That's a different company than the one Steve Jobs built.
What This Means for the Other AI Labs
The Apple Google AI deal is also a signal about how the rest of the AI market will structure itself.
For Anthropic, the lesson is that direct consumer distribution is brutally hard. Anthropic does not have a meaningful consumer device presence, and the Apple-Google deal closes off the most valuable distribution channel in consumer technology. Anthropic's enterprise focus — Claude in Slack, Claude in Notion, Claude in dev tools — is the right strategic response, but it leaves Anthropic structurally smaller in the consumer market than OpenAI or Google.
For OpenAI, the deal complicates the IPO narrative. The pitch has been that OpenAI is the default consumer AI brand. Losing the primary Siri integration to Google undercuts that. OpenAI's response is likely to be either a hardware play of its own (the Jony Ive-designed device project) or deeper integration with non-Apple device makers (Samsung, the broader Android ecosystem).
For xAI, the deal is largely irrelevant. xAI's Grok is positioned as the X-platform native AI and the political-right-coded alternative to consensus AI models. The consumer device integration market was never xAI's natural territory.
For Meta, the implications are subtle but interesting. Meta's Llama models are open-source and free. Meta has no consumer device distribution advantage (Quest VR is not at scale). But the open-source angle means that any device manufacturer who wants to avoid paying Google or OpenAI can build on Llama. The Apple-Google deal raises the bar for what a credible commercial AI license costs, which makes free open-source models more attractive by comparison.
The WWDC June 8 Tell
WWDC 2026 runs June 8 through June 12. Apple's developer conference is where the company will publicly show how Gemini-powered Siri actually performs in iOS 27.
The thing to watch is what Apple does not say. If Apple announces the new Siri capabilities without prominently crediting Google or Gemini, that's the company trying to obscure its dependence on a competitor's technology. If Apple's marketing materials emphasize "Apple Intelligence" and "Apple Foundation Models" without mentioning Gemini once, the Apple Google AI deal is being structured as a hidden licensing arrangement rather than a public partnership.
That obscuring matters because it tells you how Apple is positioning the deal internally. A company proud of a partnership puts the partner's logo on stage. A company embarrassed by a dependency hides it.
For investors evaluating Apple's strategic position, the WWDC presentation will be more revealing than any analyst report. Watch for: how often Gemini is mentioned, whether Google's branding appears anywhere, whether Sundar Pichai or Thomas Kurian make an appearance, and how the on-device vs. cloud AI split is described.
If Gemini is invisible at WWDC, you'll know Apple is treating the deal as a strategic embarrassment to be managed rather than a partnership to be celebrated. That tells you everything about how Apple actually sees its position in the AI race in 2026.
The Apple Google AI deal is structurally lopsided. The next question is whether Apple admits it.