Silicon Valley has a new hardware ambition, and it comes in the shape of a doughnut. OpenAI is developing a standalone consumer device — roughly hockey-puck-sized, screen-free, and priced somewhere north of $300 — that the company intends to bring to market by 2027. The collaborator on the industrial design is Jony Ive, the former Apple design chief who gave the world the iMac, the iPod, and the iPhone. That pedigree is doing a lot of heavy lifting in the early buzz, and OpenAI appears to be leaning into it fully.

What makes the device unusual — and, depending on your tolerance for anthropomorphized gadgetry, either fascinating or unsettling — is that it will have moving parts. According to details that have circulated from people familiar with the project, the unit will incorporate lights and sensors alongside whatever mechanical motion the design team has settled on. The effect, by design, is intended to feel less like a speaker sitting on a shelf and more like something alive. OpenAI's own framing has reportedly used the word "living" to describe it internally. That is a marketing concept, not a technical specification, but it signals exactly how the company wants consumers to relate to the thing.

The device is screen-free by intention — a pointed departure from the direction Amazon and Google have pushed their own smart home hardware, both of which have steadily added displays to their flagship products. OpenAI's argument, implicit in this design choice, is that the screen is the wrong interface for AI. The bet is that voice, motion, and ambient presence are a more natural surface for a model like GPT-4o than yet another rectangle demanding visual attention.

That framing puts this squarely in the territory of what the tech industry has lately taken to calling "AI-first computing" — the idea that the next platform shift is not a new form-factor of existing device, but a fundamentally different relationship between a user and a model. OpenAI has watched competitors and partners alike attempt versions of this. The Humane AI Pin and the Rabbit R1 both launched in 2024 as screen-free, AI-native devices; both were received by reviewers with a mixture of curiosity and disappointment, and neither found meaningful commercial traction. OpenAI is aware of this history.

What OpenAI has that those earlier entrants did not is the most recognized AI brand in consumer technology and a design partner whose track record is, by any reasonable measure, unmatched in the industry. Ive's firm, LoveFrom, has been working with OpenAI's CEO Sam Altman since at least 2023, and the acquisition of Ive's hardware startup io Products earlier this year formalized that relationship into something closer to a joint venture. The design language that emerges from that collaboration — whatever the final form — will carry credibility that a startup pitching a chest-mounted camera badge simply could not.

The price point, however, is a real variable. At $300 to $400, the device would sit well above an Amazon Echo ($100 at the high end of its lineup) and above the Google Nest Audio. OpenAI is clearly not trying to compete on commodity smart-speaker economics. The implicit positioning is closer to how Apple prices its hardware: a premium object for a user who considers themselves early on the curve and is willing to pay for the experience of being there. Whether that market is large enough to matter — or whether GPT fatigue sets in before 2027 — is a genuine open question.

There is also the matter of what, exactly, "moving parts" means for a consumer device shipped at scale. Mechanical components introduce failure modes that software does not. A speaker that squirms or pulses or physically reacts to conversation sounds compelling in a design studio; it introduces warranty and reliability considerations that a static cylinder does not. Apple's own hardware history under Ive is not without cautionary notes here — the butterfly keyboard mechanism being the most cited example of a design-forward choice that created durability problems at scale.

None of that resolves before 2027. What is clear today is that OpenAI has decided the interface wars are not over and that AI deserves a physical object purpose-built for it — not a phone app, not a chatbot window, and not a voice assistant bolted onto a commodity speaker. Whether the doughnut becomes the next iPod moment or the next expensive object gathering dust beside the Alexa nobody uses anymore will depend entirely on execution. But the ambition, at minimum, is not small.