Three years after ChatGPT launched the AI arms race, Apple has reached a conclusion most of its competitors haven't: it's not going to win this one.
According to Bloomberg's Mark Gurman, Apple has made a deliberate strategic pivot away from building competitive AI models. The company will instead position itself as the premium hardware platform through which users access AI from third parties - and take a commission on the way through.
What Got Cut
The most visible casualty is "Mulberry," Apple's codename for a standalone AI health coaching service originally planned for iOS 26. It's dead. Apple will roll out individual health features within its existing Health app instead, with health queries folded into a revamped Siri starting with iOS 27.
The conversational Siri overhaul, promised since 2025, has been delayed again to iOS 26.4 (spring 2026). The delays were partly attributed to indecision within the AI team.
Gurman stated directly that "the opportunity has passed" for Apple Intelligence to become a revenue driver. It will not be a paid product. Instead, it becomes a feature that helps sell iPhones.
The New Strategy: Be the App Store for AI
Starting with iOS 27, expected to be announced at WWDC on June 8, Siri will support third-party AI chatbots including Google Gemini, Anthropic Claude, Microsoft Copilot, Perplexity, Amazon Alexa, xAI Grok, and Meta AI alongside the existing ChatGPT integration. Apple has committed to delivering this before the end of 2026.
The business model is familiar: Apple earns App Store commissions on AI subscriptions rather than building competing products. It's the same playbook Apple used with music streaming and video - provide the platform, let others compete on it, collect rent.
The Reorg Behind the Pivot
This isn't just a product shift. Apple's AI leadership has been overhauled. John Giannandrea, SVP of Machine Learning and AI Strategy, announced his retirement after the Siri delays. Amar Subramanya, a former Microsoft AI researcher, was hired as VP of AI reporting to Craig Federighi. Federighi's software engineering group absorbed much of the AI organization. Eddy Cue took over health oversight and AI-related teams including Search and Knowledge.
The move is pragmatic. Apple tried to build competitive AI and couldn't ship on time. Its leadership reportedly concluded the company is "unlikely to become a leader in the AI race." Rather than pour more billions into a losing position, Apple is playing to its actual strength: 2 billion active devices and a captive ecosystem where it controls distribution.
For users, this could actually be good news. Instead of being locked into whatever Apple builds, you'll get to choose between ChatGPT, Claude, Gemini, and others directly through Siri. The competitive pressure between those providers should produce better results than any single vendor's assistant.
The risk is that Apple becomes dependent on companies it doesn't control for a core device experience. If OpenAI or Google decides to favor Android, Apple has limited leverage. But that's a problem for 2028. For now, Apple is doing what it does best: selling hardware and taking a cut of everything that runs on it.