OpenAI is working on a smartphone that would replace traditional apps with AI agents, according to a research note from veteran analyst Ming-Chi Kuo published Monday. The device, reportedly being co-developed with MediaTek, Qualcomm and Luxshare, would aim for mass production in 2028 and represents the company's most ambitious move yet beyond a chat window.
The news, first reported by TechCrunch on April 27, 2026, lands as OpenAI is racing to extend its reach past the nearly one billion weekly active ChatGPT users it serves through software. A dedicated hardware platform would put the company in direct competition with Apple and Google in the most lucrative consumer-electronics category in the world.
Agents instead of apps
Kuo's note describes a device designed for continuous contextual understanding of the user, with an architecture that combines on-device and cloud-based models to handle varying request types. Instead of opening discrete applications for messaging, navigation or shopping, the phone would route requests to AI agents capable of completing tasks across system features.
The analyst frames the approach as a way for OpenAI to bypass Apple and Google's app-store restrictions, which have long constrained how deeply third-party AI assistants can hook into a phone's core functions. By owning the hardware layer, OpenAI would no longer need to negotiate API access or default-assistant placement.
Supply chain coming together
MediaTek and Qualcomm are tapped for chip development, while China's Luxshare — best known as a major Apple iPhone assembler — is named for co-design and manufacturing. Kuo expects component specifications and supplier selection to be finalized by the end of 2026 or in the first quarter of 2027.
That timeline pushes any consumer launch well into 2028 at the earliest, a striking gap given the pace of AI software releases. The cadence is consistent with first-generation phone programs from new entrants, which typically take 18 to 24 months from spec freeze to retail.
Bigger hardware play
The smartphone is not OpenAI's first hardware project. The company previously announced plans to launch its inaugural hardware product in the second half of 2026, with multiple reports suggesting earbuds as the initial device. That work has been broadly linked to OpenAI's high-profile hardware collaboration with former Apple design chief Jony Ive, though Kuo's latest note does not address Ive's involvement in the phone program specifically.
OpenAI declined to comment on the smartphone rumors. The disclosure, however, is consistent with internal signals from the company that the future of consumer AI requires a device that is built around agents from the silicon up rather than retrofitted onto a smartphone OS designed in 2007.
Why it matters
If OpenAI ships an agent-first phone, it would be the first credible attempt by a software-native AI lab to dislodge the iOS–Android duopoly. The risk is enormous: hardware launches typically burn cash for years, and the device category has resisted disruption for more than a decade. But with ChatGPT now embedded in daily routines for hundreds of millions of users, OpenAI has the rarest asset in consumer electronics — a software brand strong enough to make people consider switching phones.
