2 min read AI-generated

OpenAI's First Device: a Donut Speaker, Not a Smartphone

Copy article as Markdown

New reports sharpen the picture of OpenAI's first hardware product: a portable, donut-shaped AI speaker about the size of a hockey puck, designed with Jony Ive. With cameras, moving parts, and a price around 300 to 400 dollars — planned for 2027.

Featured image for "OpenAI's First Device: a Donut Speaker, Not a Smartphone"

People have been speculating about OpenAI’s hardware plans for months. Now new reports — kicked off by Bloomberg — offer far more concrete details. And they’re surprising: the first device apparently isn’t a smartphone, but a portable, donut-shaped speaker roughly the size of a hockey puck.

What we know so far

The device is being built in collaboration with Jony Ive, Apple’s former design chief. The description sounds suitably ambitious: high-quality metal finishes, moving parts, camera sensors, dynamic lighting, and integrated speaker grills. Battery-powered and portable — you’re meant to carry it from room to room, on the nightstand, in the kitchen, or in your hand during a conversation.

The price is said to land in the $300 to $400 range, with a launch targeted for 2027. OpenAI positions the whole thing not as a conventional voice assistant but as an “AI-first computer” — a device that holds natural, ongoing conversations, much like ChatGPT’s Voice Mode today.

From smartphone to donut

The twist is interesting. Earlier reports pointed more toward a smartphone-like device. The new details correct that picture: not a screen-first phone, but a voice-centric companion without the usual form factors. That fits Ive’s stated skepticism toward displays — the idea of stepping away from the screen-bound world runs through his earlier remarks.

My take

A device that watches you, with cameras and dynamic lighting — that’s bold and tricky at once. The appeal is obvious: AI present in the room without you reaching for a phone. So is the flip side: a microphone plus a camera on your nightstand raises real privacy questions, and OpenAI will have to answer them before something like this lands in many living rooms.

For us here, the direction is what’s most interesting. While Anthropic embeds Claude deep into software, code, and workflows, OpenAI is attempting the leap into the physical world — its own device, its own presence. Two very different bets on where AI shows up in everyday life. 2027 is a long way off, and plenty of this is still rumor. But the outline is getting sharper.


Sources: