OpenAI Is Building a Phone. The Smartphone Era Is About to Get Rewritten.

Not a gadget. Not a wearable. A full smartphone. Mass production targeted for 2028.

I have been saying for a while now that the next big step for smartphones is going to be a device where the OS is an AI system. Something that can make apps for you on the fly that exactly fit your life.

The sandbox problem

Right now, ChatGPT runs on your iPhone inside Apple's permission sandbox. To do something as simple as help you order food, it has to navigate Apple's restrictions, hand off between apps, and ask for permissions it may not get. It is a guest in someone else's house.

Building their own phone means starting from the lowest layer. The AI can call whatever it needs, access your real-time context (location, activity, communication) without restrictions. No permission dialogs. No handoffs. No sandbox.

Analyst Ming-Chi Kuo frames it this way: your smartphone is the only device that captures your full real-time state at all times. That context is the most valuable input for an AI agent that is supposed to act on your behalf. Without owning the hardware, you cannot own the experience. Kuo's framing in his note is direct: "Only by fully controlling both the operating system and hardware can OpenAI deliver a comprehensive AI agent service."

Lighter tasks will be handled on-device for speed and privacy, with heavier reasoning offloaded to the cloud. All running on custom chips optimized for always-on context awareness and power efficiency, not raw speed. Kuo describes the interface as a "task stream" rather than the familiar grid of app icons. Apps, in this model, are not the unit of work. Intent is.

A strategic reversal

This is also a notable strategic reversal. OpenAI spent $6.5 billion acquiring Jony Ive's hardware startup io in May 2025, with plans for a smart speaker, glasses, and a pocket device. All non-phone form factors. The first product was supposed to ship in 2026 and explicitly avoid the smartphone category.

A smartphone changes the scope of that bet entirely. The io acquisition gave OpenAI the design DNA. Kuo's report says they are now applying it to the device category they originally said they would not touch.

Sam Altman posted on X yesterday: "feels like a good time to seriously rethink how operating systems and user interfaces are designed."

That is not a subtle hint. That is a product roadmap announcement dressed as a tweet.

The supply chain story

The most interesting part of the report is the supply chain.

Chip partners: MediaTek and Qualcomm. Kuo's note has the two chip companies collaborating on the processor design for OpenAI's custom silicon. Qualcomm shares jumped roughly 12% on the news. That is the market pricing in two things at once: the prestige of being inside an OpenAI device, and the fact that Qualcomm now has a credible answer to investors who have been asking what happens when Apple finishes building its own modems and stops needing them. MediaTek's involvement is the quieter half of the headline. They have been climbing into the premium tier of mobile silicon for years, and an OpenAI chip co-design is the kind of contract that confirms they have arrived.

Manufacturing partner: Luxshare. This is the detail that should make supply-chain watchers sit up. Luxshare is the Chinese contract manufacturer that has been steadily challenging Foxconn inside Apple's supply chain for years. They started with AirPods, moved into iPhone assembly, and have been the Apple-anointed second source for high-end devices. Picking Luxshare instead of Foxconn is a signal. OpenAI is going with the manufacturer that has the most to gain by becoming someone else's flagship partner, and the one whose roadmap is least entangled with Apple's.

Final specs lock in late 2026 or Q1 2027. That is the window where this either becomes real or quietly disappears. Things to watch between now and then: which carriers OpenAI talks to, whether they file FCC paperwork under their own name or a partner's, and whether Foxconn lands any consolation work or gets cut out entirely. The supply chain has to commit billions of dollars in capacity decisions long before the device is announced. Those decisions are where you will see the truth before any keynote.

What this means for your business

Three things if you are building software.

1. The mobile assumption is becoming negotiable. For 15 years, "mobile" has meant iOS and Android. If a third platform shows up that is genuinely AI-native, your distribution math changes. Companies that rely on app-store funnels should start thinking about what their product looks like when there is no app store.

2. Integrations are the moat, not the UI. If the future of the phone is an agent that calls services on your behalf, the businesses that win are the ones whose APIs the agent can call cleanly. Brittle integrations and login-walled flows will lose to clean, well-documented endpoints.

3. Watch the silicon news, not the keynote. Qualcomm, MediaTek, Luxshare, and the analysts who track them will tell the real story over the next 12 months. By the time OpenAI does a launch event, the strategic decisions will be locked in. The interesting reads between now and Q1 2027 are earnings calls, not press releases.

2028 is not that far away.

---

*Raptor Tech builds custom software and AI systems for businesses that want to be ready for what is coming, not surprised by it. If you are thinking about how to position your product for an agent-native future, book a free consultation or call (561) 786-7926.*

OpenAI Is Building a Phone. The Smartphone Era Is About to Get Rewritten.