OpenAI is developing an AI agent smartphone with Qualcomm and MediaTek, targeting 300-400M annual shipments by 2028.

OpenAI is building a smartphone built around AI agents rather than applications, Qualcomm and MediaTek jointly designed a custom processor and Luxshare Precision Industry’s co-design and device customization, according to Ming-Chi Kuo, an analyst at TF International Securities whose expertise in the Apple supply-chain has made him the most followed hardware analyst in the industry. Kuo plans to ship 300 to 400 million units a year if the device is successful, a number that could exceed the capacity of Apple’s iPhone and put the phone in direct competition with two companies that control about 40% of the global smartphone market. Specifications and supplier lists are expected to be finalized in late 2026 or the first quarter of 2027, with mass production targeted for 2028. Qualcomm shares rose as much as 13% in premarket trading on the report. None of the three companies, Qualcomm, OpenAI, or MediaTek, have confirmed the partnership. This is an analyst report, not an announcement, but the supply chain Kuo describes is not speculation. It’s a supply chain that already builds the devices you own.

The concept

The phone described by Kuo is not a smartphone with an AI assistant. It is a device where the AI ​​agent is the interface and the operating system is obsolete. Instead of downloading apps and navigating screens, users will interact with agents that handle tasks directly: ordering transportation, booking restaurants, managing email, doing research, writing messages. The architecture will process simple tasks on the device, including context awareness, memory management, and small AI models, while offloading complex navigation to the cloud. The device will maintain what Kuo calls a “full real-time state,” continuously capturing the user’s location, activity, interactions, and environmental context to feed agents. This is the vision that Qualcomm CEO Cristiano Amon has been putting forward throughout 2026: that AI agents will replace the mobile operating system and applications as the main layer of interaction, and that hardware must be designed from the ground up to support a continuous, energy-saving AI endpoint rather than retrofitting existing chipsets with closed neural processing units.

The idea is different from another OpenAI hardware project with Jony Ive, a former Apple design executive whose company io is building a device that is not a phone, reportedly a smart speaker with a camera first, then glasses, a light, and earbuds, with the first product expected in early 2027. OpenAI is pursuing two complementary hardware strategies: a device that rethinks what a personal phone screen looks like, but keeps everything looking like a personal computer, and returns the phone screen that runs on top of it. Apple is testing AI smart glasses with a custom chip, cameras, and Siri powered Gemini model, targeted for 2027. The question of whether AI lives on your phone, on your face, or in the speaker at your counter is being answered at the same time by every major tech company, each with a different bet. OpenAI is betting on all of them at once.

The supply chain

The credibility of the report depends on the supply chain, not the concept. Luxshare Precision Industry is a major Apple supplier that includes AirPods, Apple Watch parts, and a growing share of iPhones. Qualcomm’s Snapdragon 8 Elite Gen 5 powers 75% of the Samsung Galaxy S26 series and, for the first time, surpasses Apple in multi-core and GPU performance. MediaTek’s Dimensity 9500 matches Qualcomm and Apple in CPU performance at a lower cost with better performance. These are not feature phone providers. They are the supplier of hundreds of millions of shipped phones. Qualcomm’s acquisition of Edge Impulse, an edge AI developer platform, in 2025 demonstrated the company’s strategic commitment to on-device AI insights across device categories. The Snapdragon 8 Elite Gen 5’s Hexagon NPU delivers AI processing 37% faster than its predecessor, supports agent AI that learns from user behavior, and integrates a personal information graph and continuous context awareness through an advanced sensor hub. Qualcomm is reportedly developing custom 3D DRAM specifically optimized for AI workloads in mobile devices. The silicon of the wire that Kuo describes does not need to be invented. Parts are available. The question is whether the software paradigm works.

The financial context is important. Qualcomm stock was trading at $149.84 before the report, down from a 52-week high of $205.95, with earnings growth down 46.9% and net income down 55.1%. The company reports earnings on April 29, two days after Kuo’s report. In February, Bloomberg reported that Qualcomm gave “a tepid forecast that points to a faltering smartphone market.” The OpenAI partnership will represent a new revenue stream in a market where Qualcomm’s traditional business, which supplies modems and processors to phone manufacturers, is under pressure from Apple’s efforts to build its own modem chips and MediaTek’s entry into the Android flagship segment. Qualcomm will help build a device designed to challenge the iPhone while continuing to supply Apple with modem chips through at least 2027, a business relationship that includes a conflict in the semiconductor supply chain.

Cemetery

The AI ​​device category has produced more failures than products. The Humane AI pin, a $699 wearable with a laser projector that beamed information into the user’s palm, was bricked forever on February 28, 2025, when HP acquired Humane’s remains for $116 million and shut down the servers. The Rabbit R1, a $199 “action model” device, attracted 100,000 pre-orders but retained only 5,000 active users after five months, a 95% abandonment rate. Its founder admitted that the device was launched too early. Both failed for the same reason: they created new form factors that didn’t solve a problem the smartphone didn’t solve, at prices that required the user to carry a second device. The OpenAI call takes a very different approach. It is not an additional device. It replaces the device that 4.7 billion people already own, in the same way, with the same basic capability, but with a very different interaction model. Whether that’s enough to avoid the graveyard depends on whether agents can do what apps do, better, faster, and without the friction of learning a new paradigm.

AI is already reshaping the mobile app ecosystem, with “vibe-coded” apps flooding the App Store in such volume that Apple has had to refuse shipments. The EU is preparing to force Google to open up Android to compete with AI assistants including ChatGPT and Claude under the Digital Markets Act, which requires equal system-level access for voice and deep integration. The smartphone software landscape is already evolving. Samsung’s Galaxy S26 uses a triple AI engine with Gemini, Perplexity, and Bixby. Google’s Pixel 10 removes multi-step tasks from background AI agents. Apple Intelligence processes queries on the device with an emphasis on privacy. Every major phone manufacturer is moving to an AI-first experience, but all are hindered by backward compatibility with the billions of existing apps and apps that use them. The advantage of OpenAI, if the phone happens, is that it has no legacy. It can design a clean interaction model without worrying about whether the Instagram notification system is working or whether the banking application is working properly. The downside is that users may not want a clean slate. They may want their own apps and an AI assistant working alongside them, which is what Samsung, Google, and Apple already offer.

Question

Kuo’s prediction of 300 to 400 million shipments per year will make the OpenAI phone one of the most successful consumer products in history. For context, Apple ships about 230 million iPhones a year. Samsung ships about 220 million Galaxy phones. A newcomer reaching those volumes is unprecedented in the smartphone era. The speculation reflects the magnitude of OpenAI’s ambition, not a logical basis for a first-generation device from a company that has never produced hardware, sold through carriers, managed warranty claims, or operated a consumer-scale supply chain. Jony Ive’s device carries the same risk: a company whose expertise is in large language models is trying to become a consumer electronics manufacturer, a transition that requires skills in industrial design, supply chain management, retail distribution, and after-sales service that OpenAI does not have and cannot obtain by hiring a single designer, however talented.

The 2028 timeline gives OpenAI two years to finalize the specification, provide a secure component, build production capacity, develop the first agent software platform, negotiate carrier partnerships, establish retail distribution, and point hundreds of millions of consumers to ditch their iPhones for Galaxy phones built by the company that never shipped hardware. The Humane AI pin took longer than that and shipped the device for nine months before it was permanently disabled. Ambition is rare. The supply chain is reliable. The concept faces the real limit of the architecture of current smartphones, which were designed around applications in 2007 and have not changed since then. But the distance between a reliable report of the supply chain and the shipping product that removes the iPhone is the distance between thesis and business, and every company in the graveyard of the AI ​​device also had an idea.