OpenAI eyes AI phone

OpenAI may be doing the obvious next thing — and the much harder one. ChatGPT already lives on phones, but the company now looks serious about building hardware of its own, not just software that rides on Apple and Android. That matters because the jump from app to device changes the whole game. This week, the story moved from vague ambition to something closer to a roadmap, with new reporting around an AI-first phone and a 2027 production target. (bloomberg.com) ### Where did this come from? The hardware push stopped being theoretical on May 21, 2025, when OpenAI said io Products was merging into the company and Jony Ive would take on deep design responsibilities across OpenAI. io was not a side project. It was built by Ive with Scott Cannon, Evans Hankey, and Tang Tan, an(bloomberg.com)$6.5 billion in stock — OpenAI’s biggest acquisition. (openai.com) ### So is this definitely a phone? Not definitely. That is the first thing to keep straight. OpenAI has officially confirmed the hardware team and the ambition, but not a smartphone product. The “AI phone” framing comes from recent supply-chain and analyst reporting, especially Bloomberg’s br(openai.com)ed direction, not an announced product. (bloomberg.com) ### Why would OpenAI want its own device? Because apps have ceilings. If ChatGPT is just another icon on someone else’s home screen, Apple and Google still control the defaults, the sensors, the notifications, the payments, and a lot of the user relationship. A device lets OpenAI shape the whole interaction model —(bloomberg.com) and tighter on-device AI features. That is basically the same reason big platform companies always end up wanting hardware. (openai.com) ### Why is a phone the hard version? Because phones are brutal. A screenless gadget can dodge some expectations. A phone cannot. It has to handle radios, batteries, thermals, cameras, carriers, app ecosystems, repair, regulation, and global manufacturing at scale. The catch is that consumers (openai.com)omplete or it looks unfinished. That is why the rumored MediaTek tie-up matters. It hints that this is moving into real component planning, not just industrial-design sketches. (macrumors.com) ### What is the reported timeline? The newest detail is speed. Kuo’s latest note, echoed across multiple outlets on May 5, said OpenAI appears to be fast-tracking development, with mass production potentially starting in the first half of 2027. That is earlier than some previous expectations that pointed to 2028. The same reports (macrumors.com)ill unconfirmed by OpenAI, but they are more concrete than the earlier “maybe someday” phase. (macrumors.com) ### What would make an AI phone different? The pitch seems to be an agent-first device. Instead of opening apps one by one, you would ask the system to do things across services — book, message, summarize, compare, buy, navigate. Think less “smartphone with AI features” and more “AI assistant that happens to be a phone.” But that (macrumors.com)it got attention, then ran into reality. People say they want a new interface, but most still love screens when they actually need to get things done. (bloomberg.com) ### Why does this matter beyond one gadget? Because if OpenAI pulls this off, it stops being just a model company and becomes a platform company. That threatens Apple, pressures Google, and forces the whole industry to answer a basic question — is AI a feature inside the phone, or the thing the phone is built aroun(bloomberg.com) a phone — it has not. The news is that the company’s hardware ambitions now look organized, funded, staffed, and maybe scheduled. That makes the idea harder to dismiss. (openai.com)