Apple's Siri Overhaul Is Built on Google Gemini, and That Should Tell You Everything

Apple is reportedly rebuilding Siri from the ground up using Google Gemini technology, and I think this tells us more about the state of Apple's AI efforts than any WWDC keynote ever could.

According to Bloomberg, the next iPhone, iPad, and Mac software updates will feature a completely revamped Siri interface alongside a new chatbot-style app. The rebuilt voice assistant will include the ability to understand personal data, analyze on-screen content, and perform AI-powered web search. All of this will run on Google's Gemini model.

Look, I've seen enough spec sheets and partnership announcements to know when a company is admitting defeat without actually saying the words. This is one of those moments.

What Apple is actually admitting here

Apple has spent the better part of a decade building its own machine learning infrastructure. The Neural Engine first appeared in the A11 Bionic back in 2017. The company has acquired dozens of AI startups. It employs thousands of ML researchers. And yet, when it came time to actually compete with ChatGPT and Google Assistant's new capabilities, Apple picked up the phone and called Mountain View.

From my time in hardware, I learned that you outsource components when you can't make them yourself at competitive quality or cost. Apple outsources modem chips to Qualcomm (though it's been trying to change that for years). It outsources display panels to Samsung and LG. These are commodity components where Apple's design expertise doesn't provide a meaningful edge.

But the AI model powering your flagship voice assistant isn't a commodity component. It's the brain. And Apple just admitted its brain isn't good enough.

The vertical integration question

Apple's entire competitive moat has been built on controlling the full stack. Hardware, software, services, silicon. The company that designs its own chips, writes its own operating systems, and manufactures its own displays is now... licensing its AI from a direct competitor in the smartphone market.

This creates some genuinely awkward dynamics. Google now has insight into how Apple plans to deploy AI features across its ecosystem. Every query that runs through Gemini is, in some form, data that flows through Google's infrastructure. The privacy implications here remain unclear, and I'd expect Apple to address this at WWDC, but the fundamental tension doesn't go away.

There's also the question of differentiation. If Apple's Siri runs on the same underlying model as Google Assistant, what exactly makes the Apple experience unique? The interface? The integration with Apple services? That's a much thinner moat than having a genuinely superior AI.

What we don't know yet

It's too early to say whether this is a temporary licensing deal while Apple catches up, or a longer-term strategic partnership. Bloomberg's reporting suggests this is tied to the upcoming WWDC announcements, but the duration and financial terms of the Gemini arrangement weren't disclosed.

We also don't know how much of Gemini Apple is actually using. Is this the full model? A fine-tuned version? A smaller distilled variant optimized for on-device inference? The technical details matter here, and we basically have none of them.

I'd also note that Apple has been rumored to be in talks with multiple AI providers over the past year, including OpenAI. The fact that they landed on Google suggests either Google offered better terms, better technology, or both. But that's speculation on my part.

The bigger picture

This Siri news dropped the same day as some other significant AI-adjacent moves in the market. Snowflake jumped on a $6 billion multiyear cloud agreement with Amazon, its biggest single-day gain since 2020. Meta announced paid chatbot subscriptions to offset AI infrastructure costs. The common thread across all of these stories is that AI capability is expensive, and companies are scrambling to figure out who pays for it.

Apple has the cash to build its own models. It has the talent. It has the hardware expertise. The fact that it chose to license from Google anyway suggests that the gap between Apple's internal AI efforts and the frontier models is wider than most analysts assumed.

That's an ambitious timeline to close, especially when Google, OpenAI, and Anthropic are all iterating faster than ever. Apple might catch up eventually. But for now, the company that once told us it was years ahead on AI is shipping Google's homework with an Apple logo on it.

The real test will be what Apple announces at WWDC about its own model development. If this Gemini partnership is a bridge while Apple's in-house efforts mature, that's one story. If it's the new normal, that's a very different one.