OpenAI is going to make smartphones, mass production in 2028

Today, Tianfeng International Securities analyst Guo Mingchi’s latest industry survey indicates that OpenAI is collaborating with MediaTek and Qualcomm to develop smartphone processors, with Luxshare Precision exclusively responsible for system integration design and manufacturing, expected to mass produce by 2028.

This supply chain information may seem like just a hardware OEM update, but it actually reveals OpenAI’s complete timeline for self-developing end devices. It’s not just about creating a better ChatGPT app, but about redefining smartphones themselves—from chips and operating systems to interaction paradigms.

Why It Must Be a Phone: Three Uncompromising Underlying Logics

Guo Mingchi’s analysis presents three core reasons for OpenAI’s entry into smartphones, fundamentally pointing to the same judgment: The ultimate form of AI Agent cannot parasitize on someone else’s operating system.

First, only by fully controlling the operating system and hardware can a truly comprehensive AI Agent service be delivered.

Even if ChatGPT exists as an app on the iPhone today, it is always constrained by Apple’s permission sandbox, with a long and broken chain of user intent transmission; only by defining the device from the bottom up can AI freely invoke system capabilities and schedule hardware resources without obstruction.

Second, smartphones are the only devices capable of continuously collecting complete real-time user states.

Your phone knows where you are right now, what meetings you have next, how long you slept last night, what you recently bought, and who you are chatting with. This real-time information is the most critical material for AI to understand you. Without this link, any cloud-based large model can only be an “invisible” remote advisor.

Third, in the foreseeable future, smartphones will remain the largest category of smart terminals globally, with speakers, glasses, and earphones unable to match this scale. If OpenAI wants to establish a sustainable entry monopoly in the consumer AI track, the smartphone is an unavoidable commanding height.

Concept Design Comparison: From “App Shelf” to “Intent Engine”

Guo Mingchi also released a conceptual interface design for OpenAI’s phone, forming a stark contrast with the current iPhone home screen.

The traditional smartphone interaction core is the “App icon matrix,” where users need to recognize, open, and operate each individual app—essentially a human adapting to machine organization method.

In contrast, OpenAI’s phone will invert this logic—users’ starting point is no longer “open a certain app,” but directly expressing their needs, with the AI Agent automatically decomposing tasks, invoking underlying capabilities, and completing closed-loop execution.

This means the app as an independent entry point will be significantly weakened, relegated to a silent module within the Agent’s scheduling chain.

Technologically, OpenAI adopts a highly collaborative architecture between cloud and edge: the device’s processor continuously understands user context, manages power consumption, optimizes memory hierarchy, and performs local inference of small basic models; complex or compute-intensive tasks are sent back to the cloud.

This “edge-cloud dual-brain” architecture imposes new requirements on chip design and is the reason MediaTek and Qualcomm are engaged in deep collaboration rather than just simple supply relationships.

Reshaping the Supply Chain Landscape: Strategic Positioning of MediaTek, Qualcomm, and Luxshare Precision

From an industry chain perspective, the real weight of this news lies in “who will play a core role in the next-generation smart terminal.”

MediaTek and Qualcomm are not merely component suppliers but are co-developers of processor specifications. Final specifications and supplier allocations are expected to be finalized between late 2026 and the first quarter of 2027.

Guo Mingchi provides a highly valuable scale estimate: referencing MediaTek’s collaboration with Google on TPU Zebrafish, the revenue contribution of a single high-end AI chip is roughly equivalent to 30 to 40 AI Agent smartphone processors; meanwhile, the global high-end smartphone market targeted by OpenAI initially ships about 300 to 400 million units annually. Even with conservative penetration estimates, the replacement demand driven by upgrade cycles could constitute a long-term incremental logic at the level of chipmaker earnings reports.

For Luxshare Precision, the strategic significance of this project is even deeper.

Guo Mingchi straightforwardly states that, regardless of how Luxshare expands within Apple’s supply chain, it will be difficult to shake Foxconn’s (Hon Hai’s) core assembly position in the short term.

The exclusive design and manufacturing contract for OpenAI’s phone essentially grants Luxshare a “ticket to become the next major manufacturer”—binding top-tier clients at an early stage of a new ecosystem, which means gaining a first-mover advantage in future capacity allocation, process standards, and even overall device definition.

This “position outside Apple’s ecosystem” is precisely why Luxshare is willing to invest heavily.

Business Model Speculation: Subscription Bundling and Developer Ecosystem Closed Loop

Beyond hardware, OpenAI’s commercial path is also worth predicting.

Guo Mingchi believes OpenAI is highly likely to bundle subscription services with hardware sales—for example, offering ChatGPT Plus benefits with device purchase, or providing hardware subsidies in exchange for subscription memberships.

The deep logic behind this model is that OpenAI’s true moat is not just the gross profit of individual devices, but locking in user access through hardware, then achieving long-term monetization via continuous subscriptions and agent ecosystem commissions.

This mirrors Apple’s classic “hardware profit + App Store ecosystem” playbook, but with “applications” replaced by “agent-dispatchable capability modules.” The key metrics for developer competition will shift from “download volume” to “frequency and quality of agent invocation.”

OpenAI’s inherent advantages in this layout include: top-tier global consumer brand recognition, years of accumulated user behavior data, and industry-leading model capabilities; the highly mature smartphone supply chain means it doesn’t need to reinvent the wheel—just precisely define product specifications and deeply integrate the industry chain.

If this device launches as scheduled, its impact will go beyond “another smartphone manufacturer entering the competition”—it will be a paradigm shift in interaction: from “people searching for apps” to “directly reaching intent,” from “application ecosystem” to “agent dispatch network.”

Guo Mingchi’s supply chain information has already proven that OpenAI is not just making empty promises but is advancing this project with tangible wafer factory capacity and manufacturing contracts.

For MediaTek, Qualcomm, and Luxshare Precision, 2028 will not only be a smartphone mass production milestone but also potentially the start of a multi-year replacement cycle.