OpenAI Building Smartphone Chip With Qualcomm and MediaTek

Analyst Ming-Chi Kuo of TF International Securities says that OpenAI is working on a custom smartphone chip with Qualcomm and MediaTek. Luxshare Precision is in charge of system co-design and manufacturing.

The goal is to start mass production in 2028. The chip specs and supplier selection should be finalised by the end of 2026 or the beginning of 2027. The deal has not been confirmed by any of the companies listed. OpenAI did not answer calls for comments.

The market moved anyway. Qualcomm's stock rose as much as 12% during the day just because of Kuo's note, almost making up for all of its 2026 losses at its peak before falling back in the hours that followed.

The Vision: Kill Apps, Replace Them With AI Agents

The thing Kuo talks about is not a smartphone that has ChatGPT loaded. The design is much more radical: get rid of the app model completely and replace it with an AI agent that does things for the user. "Only by fully controlling both the operating system and hardware can OpenAI deliver a comprehensive AI agent service," Kuo said.

The technical model uses both on-device inference and cloud computing. The phone keeps track of the user's current situation in real time by managing the memory hierarchy, gathering the user's current state, and running local AI inference for immediate tasks while sending more complex computations to the cloud. If the result comes true, it will make the current app environment structurally useless. Users wouldn't have to open Uber, Google, or individual apps anymore; the AI helper would take care of requests itself.

OpenAI's strategic thinking is a lot like Apple's, but it's used for AI. Apple's own silicon gave it a performance edge that no Android competitor has been able to match. Similarly, a chip designed specifically for ChatGPT inference would get rid of the trade-offs that come with a general-purpose Snapdragon. This would also get rid of Apple and Google's control over which AI features get access to the system.

Why Apple Should Take This Seriously

For Apple, the threat is not urgent but long-term. The company relies on Google's Gemini model to power its Siri AI improvements right now, which shows that its own AI plan isn't very strong. Apple's App Store-based business model, which makes tens of billions of dollars a year from sales, subscriptions, and developer fees, faces a major challenge if OpenAI's idea of a "app-free" device takes off. For the App Store to be valuable, apps must continue to be the main way that people interact with digital services.

The planned size makes the goal even more clear. Market figures say that 300 to 400 million units will be shipped each year if the program goes as planned. This number would make an OpenAI device one of the biggest smartphone platforms in the world right away. It's not at all clear if those projections are based on actual demand or just wishful thinking.

The Credibility Problem: Standalone AI Hardware Has Failed Every Time

The idea is interesting. But the past of executions is not. Humane stopped making the AI Pin because it was making people laugh more than it was selling. Reviewers cut Rabbit R1 to pieces. It hasn't been proven that people will give up their smartphones in favour of a gadget designed specifically for AI on a commercial level. The failure rate for this group is total and recent.

The time makes it even harder to believe. In March, Fidji Simo, who was OpenAI's chief of applications at the time, told workers that the company should stop working on "too many apps and stacks" and instead focus on coding and enterprise users. After that, OpenAI shut down the Sora video generation app for consumers, merged its OpenAI for Science division with itself, and on April 17, top executives Kevin Weil and Bill Peebles left the company.

A strategic focus reset is not the same thing as a multi-year, multi-billion-dollar smartphone hardware effort. The public has not been told how to settle the disagreements between the Simo order and the Kuo report..

OpenAI bought Jony Ive's startup io for $6.4 billion in May 2025, but the company also has its own hardware project. In the second half of 2026, a smart device that isn't a phone is expected to be released. Running two hardware programs at the same time—a wearable and a smartphone—while also refocusing on business software makes execution more difficult than it needs to be.

What This Means for Qualcomm and MediaTek

Even though the Kuo story wasn't confirmed, it caused a 12% rise in Qualcomm stock during the day, which almost made up for all of its losses in 2026. Analysts expect the company to make about $10.56 billion in sales when it reports its Q2 numbers on Wednesday. The most market-moving part of that call, no matter what the headline numbers are, will be any talk about the OpenAI relationship, whether it's a confirmation, denial, or strategic colour.

MediaTek's flagship application processor business, which has been fiercely fighting with Qualcomm in high-end Android segments, would grow a lot if it worked with Qualcomm to co-develop a chip for a program that could produce 300–400 million units a year.

The OpenAI smartphone chip story is both the most bold consumer hardware play in years and the least tested in real life. The Apple vertical integration comparison is interesting from a philosophical point of view, but the history of AI hardware is a constant warning.

The mass production deadline of 2028 gives OpenAI three years to work out the execution problems that come up between the focus-reset messaging and the hardware expansion, the wearable program and the smartphone program, and the vision of getting rid of apps versus the fact that customers don't want large-scale AI devices that don't have apps.

Qualcomm's earnings call on Wednesday is the first chance for anyone named in the story to either confirm or deny it. The 12% rise and subsequent drop show that the market is still not sure if this story is transformative or hopeful until then.