OpenAI reportedly building an AI-focused smartphone, mass production tipped for 2028

OpenAI smartphone: key development details

As per Kuo, the device will be powered by custom processors co-developed with MediaTek and Qualcomm, and assembled exclusively by Luxshare, which already has a confirmed deal with OpenAI for consumer device production and also handles iPhone assembly for Apple. Specifications and suppliers are expected to be locked in by late 2026 or early 2027, with mass production targeted for 2028.

Kuo is direct about the reasoning behind the smartphone. “Only by fully controlling both the operating system and hardware can OpenAI deliver a comprehensive AI agent service,” he wrote on X. He adds two more reasons: the smartphone is the only device that captures a user’s full real-time state, which is the most critical input for AI inference, and smartphones will remain the largest-scale device category for the foreseeable future. OpenAI’s advantages, he argues, are its consumer brand, years of accumulated user data, and leading AI models, combined with a hardware supply chain that is already mature enough to execute on.

On the business model, Kuo suggests OpenAI may bundle subscriptions with hardware and build a new developer ecosystem around AI agents, not unlike how app stores work today but built around completing tasks rather than launching applications.

This smartphone project is also separate from OpenAI’s collaboration with Apple’s former Chief Design Officer Jony Ive and his startup io, acquired for around USD 6.4 billion. That is focused on categorically new devices like smart speakers with cameras, smart glasses, AI companions designed to sit alongside a phone rather than replace it. CEO Sam Altman has described those devices as the opposite of the constant attention-grabbing nature of a smartphone. The first is product is rumoured for early 2027.

The smartphone vision

Kuo shared a concept showing what this phone could actually look like in use. On the left is a traditional iPhone home screen with app icons. On the right is what Kuo calls an Agent Task Stream. There’s no icon grid but a live feed of ongoing AI tasks with status updates, progress indicators, and next steps.

Kuo’s screen shows a Tokyo business trip being booked, a Taiwan market brief being compiled and nearly ready to send, two emails flagged for approval, a family dinner reservation confirmed, and an insurance policy renewal marked complete — all running simultaneously, updated in real time. Kuo notes the chip will need to continuously understand user context, managing power consumption, memory hierarchy, and on-device small model execution. More complex tasks will route to cloud AI.

This puts OpenAI in conceptual competition with approaches that have already been tested. The Humane AI Pin removed the screen entirely and found no audience. The Rabbit R1 built a dedicated AI device that struggled to justify its existence alongside a phone. Samsung and Apple are layering AI into existing interfaces without changing the fundamental model. OpenAI, if Kuo is right, is proposing something more structural, not a feature added to a phone, but a different idea of what a phone is for.

For anyone buying a phone today, this remains a 2028 story at the earliest. The closest real-world approximations are Samsung’s Galaxy AI suite, Google’s Gemini integration, and Apple’s on-device intelligence tools, all of which work within the existing app grid rather than replacing it. OpenAI is essentially exploiting a significant gap in the industry.