Sources: Apple has renewed discussions with OpenAI about using its technology to power some features in iOS 18; talks with Google on using Gemini remain ongoing
- The iPhone maker has also been holding discussions with Google — Negotiations center on powering AI chatbot in iOS 18 update
Context & Ripple Effects
Apple's reported return to OpenAI follows its earlier exploration of Gemini for new iPhone capabilities, indicating that the company was evaluating more than one external model supplier rather than committing early to a single partner. The talks matter because they place generative AI inside the operating-system layer, where Apple controls distribution and user experience.
The subsequent report that Apple was moving toward ChatGPT features while Gemini talks continued makes this moment an important marker of a multi-provider procurement strategy, not simply a one-off chatbot negotiation.
First-order effects
- Apple gains negotiating leverage and technical options for iOS 18 by keeping both OpenAI and Google in active discussions.
- OpenAI and Google are positioned to compete for prominent access to iPhone users, subject to Apple's product, privacy, and integration requirements.
Second-order effects
- The parallel talks pressure model providers to differentiate on integration terms and capability, not solely on model quality.
- Developers and chatbot providers face an OS environment in which Apple may mediate how third-party AI services reach iPhone users, rather than leaving discovery entirely to standalone apps.
Third-order effects
- If this approach persists, mobile platforms may become AI-service gatekeepers that assemble multiple foundation-model partners behind a unified device experience.
- The later revenue-sharing approach Apple was reportedly considering suggests that AI integrations could evolve from feature licensing into platform-controlled distribution and monetization relationships.
The trend: Consumer-device platforms are shifting from building every AI capability internally to orchestrating competing model providers at the operating-system layer.