Apple gave up on Siri

In March 2026, Bloomberg reported that Apple plans to open Siri to rival AI services as part of its upcoming iOS 27 update. Third-party AI apps, including Alphabet's Gemini and Anthropic's Claude, will be able to integrate directly with Siri, letting users route queries to whichever service they prefer. OpenAI's ChatGPT will no longer hold an exclusive partnership. This is not a minor product tweak. This is Apple, the company that built its empire on owning every layer of the stack, admitting that it cannot win the AI assistant race alone. And it might be the smartest thing the company has done in years.

The long, slow decline of Siri

Siri launched in October 2011 alongside the iPhone 4S. It was, by any measure, a landmark moment. The first mainstream voice assistant, it turned natural language interaction from science fiction into something you could do on your phone. Apple was years ahead of the competition. Amazon's Alexa didn't arrive until 2014. Google Assistant launched in 2016. But being first didn't translate into being best. While competitors invested heavily in natural language understanding and generative AI, Siri remained largely a command-and-control system. It could set timers, send texts, and check the weather, but it struggled with anything that required real comprehension or multi-step reasoning. By 2025, the gap had become embarrassing. ChatGPT was handling complex conversations, writing code, and analyzing documents. Google's Gemini was deeply integrated across Android and search. Siri was still occasionally confused about what month it was. As one 9to5Mac columnist put it, Siri had become "an unmitigated disaster," with its degree of dysfunction actually increasing over time rather than improving. Apple knew this. At WWDC 2025, the company conspicuously avoided launching the long-promised "more personal Siri." A CNBC report from December 2025 noted that Apple had effectively told investors not to bother asking about AI that year, with the promise that a major Siri overhaul would come in 2026. Deepwater Asset Management's Gene Munster warned that Apple needed to "deliver a 10 out of 10 when this new Siri comes out." The question was never whether Apple needed to fix Siri. The question was how.

A fundamental break from Apple's DNA

Apple's entire identity is built on vertical integration. The company designs its own chips, its own operating systems, its own applications, and increasingly its own services. Control is the product. When you buy an iPhone, you're buying into a closed ecosystem where Apple has optimized every interaction between hardware and software. Opening Siri to third-party AI services is a departure from that philosophy that's hard to overstate. Apple is essentially saying: we'll provide the interface, but the intelligence can come from somewhere else. Users will be able to ask Siri a question and have it answered by Claude, Gemini, or whatever AI service they prefer. This is not how Apple usually operates. When Apple Maps launched in 2012 as a replacement for Google Maps on iOS, it was a catastrophe. Melting bridges, misplaced landmarks, entire cities disappearing. The Washington State Department of Transportation had to tweet reassurances that the Tacoma Narrows Bridges had not, in fact, melted. Apple CEO Tim Cook issued a public apology, and the executive who oversaw the launch was eventually pushed out. But Apple's response to the Maps debacle was characteristically Apple: fix it internally. The company acquired mapping companies like HopStop, Embark, and WifiSlam. It rebuilt the product from the ground up. A decade later, Apple Maps is genuinely competitive with Google Maps. The lesson Apple took from that experience was that patience and internal investment could overcome any initial failure. With Siri and AI, Apple appears to have reached a different conclusion. The gap isn't something that can be closed with a few acquisitions and a couple of years of iteration. The AI race moves too fast, the investment required is too massive, and the leading players have too much of a head start. Rather than spending years building something that might still fall short, Apple is choosing to become the platform that gives users access to the best AI, wherever it comes from.

The platform play

Here's the thing that makes this move strategically brilliant rather than a surrender: Apple doesn't need to have the best AI. It needs to control the device that all AI runs on. There are roughly 2.2 billion active Apple devices worldwide. The iPhone remains the dominant smartphone in the world's most valuable markets. Every AI company, from OpenAI to Google to Anthropic, wants to be where the users are. And the users are holding iPhones. By opening Siri to third-party services, Apple transforms itself from a competitor in the AI race into the arena where the race takes place. Instead of spending tens of billions trying to match Google's model capabilities or OpenAI's research output, Apple can take a cut of every AI interaction that happens on its platform. The economics are already working. MacDailyNews reported that Apple is set to pocket over $1 billion from rival AI apps in 2026 through App Store commissions on AI subscriptions. That's revenue Apple earns without training a single model or running a single GPU cluster. This is the same playbook Apple used with the App Store itself. Apple didn't need to build every app. It built the platform, set the rules, and took 30% of every transaction. The AI version of this strategy is even more elegant: AI companies need Apple's distribution more than Apple needs their technology.

The distribution deal of a lifetime

For OpenAI, Google, and Anthropic, getting default or preferred access on iPhones is arguably the most valuable distribution opportunity in AI right now. Consider the stakes. Bloomberg reported in January 2026 that Apple plans to use Google's Gemini technology to underpin its own chatbot version of Siri, while also allowing competing services to plug in. That means Google gets to power the default AI experience on the world's most premium smartphone platform, while Apple retains the option to switch providers or add alternatives at any time. This creates a fascinating competitive dynamic. AI companies will compete fiercely for placement on the iPhone, potentially offering Apple favorable terms, revenue sharing, or exclusive features. Meanwhile, Apple maintains leverage by ensuring no single provider becomes indispensable. Elon Musk clearly understood the stakes. His xAI startup sued Apple and OpenAI over the original Siri-ChatGPT exclusivity deal, accusing the two companies of conspiring to ensure their continued dominance in the AI market. Musk wanted Grok on the iPhone alongside ChatGPT. With the iOS 27 changes, he may get his wish, but so will everyone else.

Samsung is already doing this

Apple isn't the only company that's figured out this playbook. Samsung has been moving in a similar direction, just with different partners. In March 2026, Samsung launched its browser for Windows with Perplexity AI integration, bringing agentic AI capabilities to desktop browsing. But the partnership goes deeper than a browser. Perplexity is now integrated at a system level into the Galaxy S26, powering search and reasoning for both the Perplexity assistant and Samsung's Bixby. It's the first time Samsung has given a non-Google company OS-level access on its phones. The AI assistant layer is fragmenting. Apple is partnering with Google for its chatbot Siri while opening the door to competitors. Samsung is partnering with Perplexity while keeping Bixby as the device control layer. The common thread is that hardware makers are separating the "intelligence" from the "interface," treating AI capability as a pluggable component rather than something they need to own outright. This fragmentation benefits consumers, who get choice, and it benefits hardware makers, who get to leverage competition among AI providers. The losers, if there are any, are AI companies that bet everything on having the best model without securing distribution.

The privacy paradox

There's an elephant in the room, and it has an Apple logo on it. Apple's entire brand identity for the past decade has been built on privacy. "What happens on your iPhone, stays on your iPhone" was not just a marketing tagline. It was a philosophical position that differentiated Apple from every other major tech company. Apple Intelligence was designed around on-device processing and Private Cloud Compute specifically to avoid sending user data to external servers. Routing Siri queries to third-party AI services fundamentally complicates that promise. When a user asks Siri a question and it gets forwarded to Claude or Gemini, that query leaves Apple's ecosystem entirely. The privacy guarantees that Apple controls, including on-device processing and encrypted cloud compute, no longer apply. The user's data is now subject to whatever privacy policies Anthropic, Google, or OpenAI have in place. Apple is clearly aware of this tension. In November 2025, the company updated its App Review Guidelines to specifically regulate apps sharing personal data with third-party AI, requiring clear disclosure and explicit user permission. The current ChatGPT integration in Siri already requires user consent before any query is sent to OpenAI. But there's a difference between requiring a permission prompt and maintaining the privacy-first brand that justifies Apple's premium pricing. As more queries flow to external AI services, the line between "Apple protects your privacy" and "Apple routes your data to whichever AI company you choose" becomes increasingly blurry. The pragmatic compromise, keeping sensitive tasks on-device while outsourcing complex queries to vetted cloud partners, is technically sound. But it's a harder story to tell than "your data never leaves your device."

Buying time or changing strategy?

There's a credible argument that Apple isn't giving up on AI at all. Instead, it's buying time. Apple has been quietly investing in on-device AI capabilities through its Apple Foundation Models framework, which gives developers access to powerful local models that run entirely on Apple silicon. The company's hardware advantage, custom chips designed specifically for machine learning workloads, gives it a genuine edge in on-device AI that no other company can easily replicate. The iOS 27 strategy might be a bridge: let third-party AI services handle the heavy lifting today while Apple builds something better for tomorrow. If Apple can eventually deliver an AI assistant that's competitive with the best cloud models but runs entirely on-device, the privacy advantage becomes overwhelming. But that's a big "if." The pace of improvement in cloud-based AI models has been relentless. Every few months, the leading models get meaningfully better at reasoning, coding, and handling complex tasks. Matching that trajectory with on-device models constrained by the thermal and power limits of a smartphone is an enormous technical challenge. More likely, the future is hybrid. Apple will maintain strong on-device capabilities for privacy-sensitive tasks while giving users the option to tap into more powerful cloud models when they need them. The question is whether Apple or the third-party providers end up controlling that boundary.

What this actually means

Apple opening Siri to rival AI services is not a white flag. It's a recognition that in the AI era, the most valuable position might not be having the best model. It might be controlling the device that 2.2 billion people use to access AI, period. The company that launched Siri in 2011 and kickstarted the voice assistant revolution spent the next 14 years watching competitors surpass it. The natural instinct would be to try to catch up, to pour billions into AI research and model training until Siri could compete head-to-head with ChatGPT. Instead, Apple is doing something more interesting. It's turning the iPhone into an AI platform, one where the best models compete for access to Apple's users rather than the other way around. It's the same strategic insight that made the App Store one of the most profitable businesses in history: you don't need to win every game if you own the stadium. Whether this works depends on execution. Apple needs to build an integration framework that's seamless enough that users actually want to use third-party AI through Siri rather than opening separate apps. It needs to maintain meaningful privacy protections even as queries flow to external services. And it needs to keep investing in its own AI capabilities so that the platform play doesn't become a permanent dependency. But the strategic logic is sound. In a world where AI capability is rapidly commoditizing, distribution is the real moat. And nobody has better distribution than the company that makes the device in your pocket.

References