OpenAI develops smartphone where AI agents will replace traditional apps

OpenAI may release its own smartphone where artificial intelligence completely displaces traditional mobile applications. According to analysts, serial production of the device is planned for 2028 — meaning active development is already underway right now. The idea is no accident.

OpenAI has long been moving toward creating its own hardware. Sam Altman's company is in negotiations with legendary designer Jony Ive — the same designer behind the visual language of iPhone and MacBook at Apple. Their joint venture, according to available information, has already received funding from SoftBank and is aimed at developing next-generation devices where AI will not be a supplementary tool, but the primary interface for interaction with the world.

OpenAI's ambitions in hardware didn't emerge yesterday. Back in 2023, the company was already negotiating the creation of a consortium for AI chip production. ChatGPT by that point had already become the fastest-growing product in the history of the technology market, and Altman understood: dependence on other people's devices is a strategic vulnerability.

Its own hardware ecosystem would allow controlling the entire user experience from start to finish. The key idea of the project is to completely replace the "one task, one application" model with an agent-based architecture. The user formulates a request in natural language, and the AI agent itself decides how to execute it: open the necessary service, place an order, find data, send a message.

No icons on the home screen, no App Store, no manual updates — just a conversation with an agent that will do everything for you. Precedents have already existed — and they failed. Humane AI Pin and Rabbit R1 came out in 2024 and never became a mass phenomenon.

The reason is simple: the agents of that time turned out to be insufficiently intelligent, and the hardware was too limited. But both gadgets appeared before GPT-4o, before Claude 3.7, before the wave of specialized agent frameworks.

Over two to three years, the gap between what agents are capable of and what users expect could drastically narrow. Models have become faster, cheaper, and significantly more reliable. The threat to Apple and Google is obvious.

Both companies control the mobile market through an ecosystem of applications: App Store and Google Play generate tens of billions of dollars annually. If agents begin to perform tasks directly, without requiring the installation of third-party programs, the business models of these platforms will come under serious pressure. It is no accident that Apple is actively integrating Apple Intelligence into iOS, and Google is betting on Gemini as the core of the Android ecosystem — both companies sense the threat and are preparing their defense.

OpenAI, in this race, does not yet have its own hardware platform. Partnerships with Samsung and Microsoft help promote ChatGPT through existing devices, but this is not the same as full control over the platform. Its own smartphone could become the entry point for hundreds of millions of users who by default would enter the OpenAI ecosystem — without intermediaries in the form of Apple or Google.

2028 is a realistic timeline, though a tight one. Developing a smartphone from scratch, passing certifications, building supply chains in Asia — all of this requires several years of sustained work. If analysts are correct, then development is already underway at full speed.

The next 12–18 months should show the first prototypes or at least official confirmation from the company itself. We are standing on the threshold of a new hardware cycle in AI. After the wave of chatbots and embedded assistants, the next battle is unfolding over a physical device that will become the main interface between humans and machine intelligence.

OpenAI clearly wants to occupy this position first — and 2028 may be the year when a smartphone ceases to be just a smartphone.