OpenAI Is Building a Phone. It Has Never Built Anything Like It.
The pressure to own the layer beneath the AI
OpenAI spent $6.4 billion on a startup with no shipped hardware. Now it is reportedly building a phone anyway.
That is the tension inside the OpenAI phone story. Last May, OpenAI acquired io Products — the hardware startup co-founded by Jony Ive, the designer behind the original iPhone — for roughly $6.4 billion in all-equity. The stated ambition was a screenless, voice-first AI companion, a device designed for ambient computing rather than the glass rectangle most people carry. Lehane, OpenAI's chief global affairs officer, said at Davos the device would arrive in the second half of 2026, with initial production of 40 to 50 million units through Foxconn.
That plan appears to have changed.
On Monday, Ming-Chi Kuo — the supply chain analyst whose record on Apple's component plans is the industry's closest thing to a weather forecast — published a revised timeline. OpenAI is now fast-tracking a full smartphone, targeting mass production in the first half of 2027, roughly a year ahead of the previous schedule. The device will run a customized MediaTek Dimensity 9600 processor built on TSMC's N2P process, with the image signal processor as what Kuo calls the "headline spec" — an enhanced HDR pipeline designed to give the phone sophisticated visual understanding of its surroundings. The chip also includes a dual-NPU architecture for handling different AI workloads simultaneously, along with LPDDR6 RAM and UFS 5.0 storage.
If development stays on track, Kuo estimates combined shipments in 2027 and 2028 could reach around 30 million units. That number deserves scrutiny. Apple's iPhone moved 3.3 million units in its first full year. OpenAI has never manufactured or distributed hardware. It has no carrier relationships in the United States, no retail footprint, and no hardware track record to point to. 30 million units in two years is not an iPhone-beat — it is a Samsung-scale ambition from a company that has never made a phone.
The shift from the io Products concept to a smartphone is the part of this story that most outlets are underweighting. OpenAI did not buy io to build a phone. The acquisition was built around an AI companion with no screen, a device designed around voice and ambient sensing. The pivot to a smartphone — the most competitive hardware category on earth, dominated by Apple and Google, with established supply chains, carrier deals, and retail networks — is not a natural evolution. It is a bet that the screenless companion did not have a viable path to the scale OpenAI needs.
Why the urgency? Kuo cites two factors: intensifying competition in AI agent phones and the potential to support a year-end IPO narrative. OpenAI has said nothing publicly about any of this. But Altman did post last week that "it feels like a good time to seriously rethink how operating systems and user interfaces are designed" — a statement that reads differently in the context of Kuo's report than it does in isolation.
The structural question underneath all of this is whether AI companies can survive as pure software. OpenAI's current business is an API that other companies build on — which means its distribution lives inside someone else's product. Apple Intelligence, Google's Gemini, and Samsung's Galaxy AI are all competing to own the assistant layer on devices OpenAI does not control. Every time a user opens ChatGPT through an iPhone, OpenAI is one decision away from being disintermediated.
Building its own hardware is one answer to that problem. It is also the most expensive and highest-risk answer. Humane tried something adjacent with the AI Pin — a device with no screen, focused on ambient AI interaction — and failed to find product-market fit despite genuine technical ambition. The smartphone monoculture is not easy to break, and the history of ambitious hardware plays by software companies is mostly a history of retreats.
What OpenAI has that Humane did not is capital, a well-known design team, and the IPO clock ticking. Whether those advantages are enough against Apple and Google — who control the distribution layer and are actively working to make their own AI assistants indispensable — is the question this product will answer, if it ships.
For now, the company has not confirmed any of this. The Kuo report is a supply chain signal, not an announcement. The timeline could slip. The specs could change. But the direction is clear: OpenAI is no longer a pure software bet.
What to watch: whether OpenAI announces the device before its rumored year-end IPO, and whether it secures carrier partnerships in the US — two things that would turn Kuo's supply chain signal into something considerably more concrete.