The partnership concluded in 2024 between Apple and OpenAI was to symbolize the American manufacturer’s official entry into the era of generative artificial intelligence. Two years later, the relationship turned into a power struggle. Bloomberg tells us last night that OpenAI is now considering legal action against Apple, accused of not having respected the spirit of the partnership and of having voluntarily limited the exposure of ChatGPT in the iOS ecosystem.
The dispute goes far beyond the contractual disagreement and sheds light on the strategic doctrine that Apple has applied for more than twenty years to all of its technological partners: integrating innovations that have become essential without ever giving up control of the interface, distribution or user relations.
Since the launch of the iPhone in 2007, Apple has built its power on a logic of vertical integration. The company accepts partners as long as they enhance the value of its devices and remain interchangeable. As soon as a player threatens to become an autonomous platform within the Apple ecosystem, the group reduces its dependence or reinternalizes the strategic function concerned.
This logic has already been demonstrated with telecom operators at the launch of the iPhone, with Adobe and Flash, with Meta during developments linked to advertising tracking, with Spotify on integrated payments, or even with Epic Games during the conflict around the App Store. Generative AI does not constitute a break in this doctrine; it is a new application.
When the partnership with OpenAI was announced at WWDC 2024, Apple was already significantly behind in conversational models and generative assistants. The integration of ChatGPT into Siri then responds to an immediate industrial need: to quickly offer credible AI capabilities while Apple’s internal models are not ready.
But Apple immediately frames this integration so that ChatGPT is neither the native engine of Siri nor a central layer of iOS. The user must often explicitly invoke the ChatGPT name to trigger certain functions. Responses appear in limited windows, and integration remains peripheral and controlled by the Apple interface.
OpenAI hoped to reproduce a pattern close to the historic partnership between Apple and Google around Search in Safari. This agreement allowed Google to become the dominant force in mobile while generating tens of billions of dollars in revenue for both companies. OpenAI anticipated comparable dynamics, with mass access to iPhone users, accelerated growth in ChatGPT subscriptions, and deep integration into the Apple software experience.
But Apple never considered such a dependence. On the contrary, the group seeks to prevent the emergence of an AI player capable of becoming indispensable inside iOS. The planned opening of iOS 27 to several artificial intelligence providers confirms this direction. Apple is already testing integrations with Anthropic’s Claude and Google’s Gemini. The goal is to transform AI models into competing, interchangeable services within an orchestration layer controlled by Apple.
This strategy aims to commoditize AI models. In this architecture, the value no longer lies solely in the quality of the model, but in the control of the interface, the terminal and the user relationship. Apple seeks to prevent OpenAI from becoming the future equivalent of Google Search: an essential layer capturing user intent and a growing share of digital revenue.
The current conflict thus reveals the shift in the center of gravity of the AI industry. For several years, the market considered that the best models would automatically end up dominating usage. Apple recalls another structural reality of the digital economy: access infrastructures remain more powerful than the technologies they host.
In mobile, Apple controls:
- hardware;
- the operating system;
- the App Store;
- distribution mechanisms;
- software integration rules;
- behavioral data linked to device use.
This position allows the group to maintain the role of orchestrator, even in the face of more technologically advanced partners in certain areas. For OpenAI, this dependence now constitutes a major strategic risk. The company understands that mastering models is no longer enough. Without control of the physical and software interface, an AI actor remains vulnerable to dominant platforms.
OpenAI’s acquisition of the company created by Jony Ive illustrates precisely this awareness. By recruiting several former Apple executives and engineers, OpenAI is looking to build its own hardware layer and user access point. For Apple, this development transforms OpenAI from a tactical ally into a potential competitor.
The real issue therefore goes beyond Siri or ChatGPT. The battle is over control of the future dominant personal computing interface. After the browser, the smartphone and social networks, the conversational assistant could become the next central layer of digital uses, and Apple refuses to leave this strategic position to an external player.
The company thus applies to AI the same doctrine that has structured its industrial history: integrate critical technologies, maintain partners in a position of dependence and maintain absolute control of access infrastructures.
This sequence comes at a time when the AI industry is entering a phase of strategic consolidation. The first cycles of experimentation focused on models give way to a much more structuring battle around infrastructures, interfaces and global distribution.
The announcement of the opening of iOS 27 to several AI providers illustrates this transition. Apple does not want to depend on a single player and seeks to transform generative models into interchangeable services within an environment that it fully controls. For OpenAI, this development constitutes a strategic warning: technological performance does not guarantee control of uses if access to users remains controlled by third-party platforms.
The conflict also reveals a broader recomposition of the market. Microsoft strengthens Copilot in Windows, Google integrates Gemini into Android and its search engine, Amazon supports Anthropic. The industry is no longer structured around a single technological champion, but around integrated blocks combining models, cloud, terminals, data and distribution.
In this context, the confrontation between Apple and OpenAI appears to be one of the first major industrial conflicts of the era of AI assistants. It marks the start of a broader battle for control of the future dominant interface in personal computing.
