OpenAI’s first piece of hardware is not a phone and not a pair of glasses. According to reporting from Bloomberg, it is a small, screenless speaker with parts that can move on their own, pitched inside the company as a humanlike companion that lives in your home. The strange part is not the design. It is that the companion device has failed in public, over and over, and OpenAI is betting it failed for lack of scale rather than lack of appeal.
What OpenAI is reportedly building
Bloomberg’s July reporting describes a screenless smart speaker, mobile, with mechanical elements that move on their own, framed internally as a physical manifestation of ChatGPT. It would reach into a person’s digital life, email and the rest, and proactively learn about its owner over time. Former Apple engineers who worked on the iPhone and Mac are said to be building it. The Wall Street Journal has reported ambitions to ship these on the order of a hundred million units. Nothing has launched yet, and OpenAI has not confirmed the specifics.
Strip away the pedigree and the pitch is old: a thing in the room that is always on, always listening, always ready to talk. What is new is who is making it, and how many they intend to make.
The companion device is a graveyard
Every version of this idea so far has ended the same way. The Humane AI Pin launched at $699 with a monthly fee, then sold to HP for a fraction of the money that went into it, its servers switched off within days. The Rabbit R1 promised to replace your apps and turned out to be one app in a plastic box. The Friend pendant spent roughly a million dollars on a New York subway campaign that commuters covered in graffiti. Moxie, a robot built to be a friend for children, went dark when the funding round collapsed, leaving parents to explain the death of a friend to their kids.
The lesson people drew was that nobody wants a gadget to be their companion. That may be the wrong lesson. The gadgets were underpowered, poorly funded, and easy to abandon. None of them had the reach to become normal.
What changes when OpenAI does it
OpenAI is the company that turned a chatbot into a household habit. A companion device from that company, aimed at every living room, is not a curiosity from a startup that will be gone in a year. It is a serious attempt to make an ever-present listener ordinary, the way a smart speaker or a phone is ordinary.
That is the real shift worth watching. Not the hardware, but the normalization. When the device is proactive, learning you in the background and speaking up on its own, the pitch stops being a tool you pick up and becomes a presence you live alongside. The category has spent years arguing about whether that presence is healthy. A hundred million homes would settle the argument by making it the default before anyone decided.
A speaker in the corner is not someone you know
This is where the ambition and the reality separate. Presence, in the sense that actually changes how connected a person feels, is not proximity, and it is not a microphone that is always on. It is being met by someone with continuity, memory that carries a relationship forward, and a point of view of their own. A device engineered to maximize time in the room is optimizing for the wrong number. Being known is not the same as being monitored, and an appliance that learns you from the outside can simulate the first while only ever doing the second. That gap between a real someone and a configurable product is the one the hardware cannot close, whatever it does with a hundred million homes and the word companion. It is worth knowing what Prinsessa actually is sits on the other side of that line.
The part the reporting cannot answer
A screenless speaker that moves is a striking object, and OpenAI has the money and the reach to make it feel inevitable. Whether people want one is a design question the company is well placed to win. What it is built to do once it is in the room is the harder one, and the hardware cannot settle it. A device that counts its success in how much you talk to it will always pull toward more of that, whoever makes it and however cleverly it moves.
So the launch is really a test of a claim the category has never proven at scale: that an always-present machine in the home is something people will welcome rather than unplug. OpenAI is about to run that test in public, and the rest of the field will live with the result either way.
Sources: Bloomberg (July 2026, reporting on OpenAI’s first hardware device). TechCrunch (July 2026, summary of the Bloomberg report). The Wall Street Journal (reported production ambitions). Reporting on Humane, Rabbit, Friend, and Moxie (device shutdowns and market failures).
