On April 27, 2026, TF International Securities analyst Ming-Chi Kuo presented industry research indicating that OpenAI, together with MediaTek and Qualcomm, is developing smartphone processors, with Luxshare positioned as the exclusive partner for joint system design and manufacturing. Mass production is targeted for 2028, with specifications and suppliers expected to be finalized between late 2026 and Q1 2027. While OpenAI has not officially announced any smartphone plans, in 2025 it integrated Jony Ive's io Products team, signaling its intent to design its own device experiences for the AI era. This observation is notable in that it suggests a possibility beyond screenless dedicated AI devices—a redesign of the smartphone itself centered on AI agents.
The 2028 Mass Production Report Views AI Devices as a Redesign, Not a Replacement, of Smartphones
In the device vision Kuo describes, rather than users opening apps one by one, an AI agent understands the user's state and goals and directly handles the necessary tasks. In processor design, power consumption, memory hierarchy management, and on-device execution of small models become important, with heavier inference handled on the cloud side. This differs from simply adding AI features to enhance a smartphone—it's a concept that integrates OS, hardware, and cloud inference as one unified design.
The 2028 mass production timeline also indicates this isn't something that will produce a product immediately. Kuo notes that specifications and supply chain details may not be finalized until late 2026 to Q1 2027, meaning what we have now is closer to an observation of design philosophy and early supply chain activity. Foreign tech media reports also explicitly note that OpenAI, Qualcomm, MediaTek, and Luxshare have not officially confirmed this smartphone plan. What's worth focusing on here isn't "confirmation of an OpenAI-made phone," but rather the shift where "discussions about dedicated AI devices have expanded from screenless devices to a redesign of the smartphone."
The lineup of supply chain players Kuo mentions also reveals the nature of this concept. MediaTek and Qualcomm are major smartphone SoC makers, and Luxshare is a manufacturing company with experience in Apple's supply chain. If OpenAI intends to optimize not just its own AI models but also device power management, memory, connectivity, sensors, and manufacturing yield, it makes more sense to leverage the robust component and manufacturing infrastructure of the existing smartphone industry. Rather than building an AI device as an entirely new gadget, riding on the smartphone industry's mature mass-production capabilities is closer to the premise needed to talk about hundreds of millions of units.
The Figure of 300–400 Million Units Reflects the Size of the Premium Market, Not a Bid to "Beat the iPhone"
Kuo stated that if the initial target is the global high-end smartphone market, annual replacement demand of 300–400 million units could become a long-term growth driver. Some media outlets have sensationalized this figure with headlines framing it as "beating the iPhone," but in reality, this doesn't mean an OpenAI-made device would capture that entire volume from the start. The benchmark to consider here is the actual market size: according to IDC's final 2025 data, global smartphone shipments totaled 1.2617 billion units, with Apple at 251.7 million units and Samsung at 241.2 million units.
The 300–400 million unit figure exceeds Apple's single-brand shipments in 2025 and represents roughly a quarter to a third of the entire global market. For a new entrant in AI devices, this is an extraordinarily large number—one that would only be reached after clearing every hurdle: mass production, pricing, carrier relationships, repair networks, app compatibility, regional certifications, and privacy protection. Kuo's intent here is closer to an investor-oriented perspective: for MediaTek and Qualcomm, smartphone-scale replacement demand represents a larger long-term opportunity than a single AI accessory. This isn't a figure indicating the probability of success as a consumer product, but rather an upper-bound hypothesis showing how large semiconductor and assembly demand could grow if AI agents truly become central to smartphone usage.
IDC's 2025 data shows that even as the smartphone market matures, it remains enormous. While total shipments that year grew only 2.0% year-over-year, both Apple and Samsung recorded growth in the 7% range, driven by premium demand and AI-related features supporting upgrade cycles. If OpenAI enters this market, the challenge isn't confirming that "the smartphone market is still large"—it's whether users accustomed to iOS and Android can be convinced that AI agent-centric operation is more convenient than existing apps. Shipment volumes will depend on whether this experience can replace daily payments, communications, photography, navigation, and work collaboration without disrupting them.
The io Integration Confirms OpenAI's Direction Toward Designing Its Own Device Experience
What OpenAI has officially confirmed is that on May 21, 2025, Jony Ive and Sam Altman jointly announced the integration of the io Products team, and that on July 9 of the same year, the team formally joined OpenAI. According to OpenAI's announcement, Jony Ive and LoveFrom would remain independent while taking on deep design and creative responsibilities across OpenAI as a whole. The announcement did not reveal specific product names or forms, but it expressed a problem awareness that existing products and interfaces have constrained the relationship between humans and AI experiences.
What can be inferred from the official announcement is that OpenAI is not just distributing models as apps, but is seeking to redesign AI usage itself, including hardware and interfaces. However, whether the first major product from this effort will actually be a smartphone, or a different category such as a screenless device or a home device, remains unconfirmed. In February 2026, MacRumors reported, based on court documents, that a Jony Ive-related OpenAI device would not ship to customers before 2027, and that same month, citing information from The Information, also reported the possibility of a smart speaker form factor. Both are peripheral reports, not evidence that OpenAI has announced a smartphone plan.
The smartphone theory matters partly because there are already examples of dedicated AI hardware failures in the market. Screenless devices or voice-centric devices can capture everyday context, but they tend to sacrifice the strengths that existing smartphones already have—confirmation, editing, payment, sharing, and app integration. If AI agents are truly meant to become entities that "handle tasks on the user's behalf," the device needs deep access to cameras, microphones, location, notifications, payments, connectivity, accounts, and app assets. Smartphones already possess this gateway, and even if OpenAI expands into other device categories, there's a certain logic to the view that it will ultimately move toward a smartphone-form redesign.
The Biggest Constraint Isn't the Chip—It's Trust in Handling Constant Context and Control Over the OS
The hardest challenge of an AI agent-type smartphone can't be solved by a dedicated processor alone. For a device to understand a user's "real-time state," it needs access to some of the most sensitive information on today's smartphones—location, schedule, messages, calls, photos, payments, and browsing activity. Even when dividing processing between on-device and cloud, deciding which information stays locked on the device, which inference gets sent to the cloud, and how to explain this to users so they can control it—these decisions will determine the product's trustworthiness. The reason OpenAI wants its own OS or deep hardware integration isn't just about performance—it's because this permission architecture is difficult to complete within the constraints of existing OSes.
The OS and app ecosystem also remain major unresolved issues when interpreting the 2028 mass production report. Whether OpenAI builds on Android, creates its own OS, or has AI agents operate existing apps will significantly change how developers, carriers, regulators, and users perceive the product. Apple's strength doesn't come solely from device polish—it comes from the integrated buildup of the App Store, iMessage, payments, peripherals, privacy messaging, and support networks. For OpenAI to create an experience capable of reaching hundreds of millions of units, what matters isn't just model performance, but a migration design that doesn't disrupt users' existing life infrastructure.
As of April 2026, OpenAI's AI smartphone is not an announced product. What can be confirmed is that Kuo has presented a 2028 mass production report involving MediaTek, Qualcomm, and Luxshare; that OpenAI has taken steps toward device design for the AI era through the io integration; and that the smartphone market still sees annual shipments exceeding 1.2 billion units. What to watch next includes signs of specification and supplier finalization expected between late 2026 and Q1 2027, official product descriptions from OpenAI, and concrete language regarding OS and privacy design. The competition for AI agent devices will become not just about joint processor development, but about who earns the right to hold users' daily context.