OpenAI has spent more than a year teasing a mysterious hardware project built with former Apple design chief Jony Ive. Now, according to a new Bloomberg report from Mark Gurman, the shape of that project has finally leaked, and it is not what most people expected. The company’s first standalone device is reportedly a doughnut-shaped smart speaker, roughly the size of a hockey puck, priced somewhere between $300 and $400.
That price puts it well above a standard Amazon Echo or Google Nest speaker, but comfortably below a flagship iPhone. It also arrives at a strange moment: OpenAI is simultaneously trying to launch this device and defend itself in an active lawsuit from Apple, which accuses the company of stealing confidential manufacturing techniques to build it. The device does not have a name yet. What it does have, according to multiple outlets now confirming pieces of the same Bloomberg report, is a genuinely odd design philosophy that treats a smart speaker less like a gadget and more like a small, slightly alive object sitting on your kitchen counter.
The Shape Nobody Predicted
Most smart speaker leaks follow a predictable pattern. Cylinder, puck, fabric mesh, maybe a screen bolted on. OpenAI’s device skips all of that and goes with a ring, essentially a doughnut you could hold in one hand. Bloomberg’s sources described it as battery-powered, which already separates it from Apple’s HomePod and most of Amazon’s Echo lineup, since those need to stay plugged into a wall.
Being cordless changes what the device is actually for. A stationary speaker sits in one spot and waits for a wake word. A battery-powered ring you can carry from the kitchen counter to the nightstand to your hand while you pace around the living room is a completely different kind of product, and that’s clearly the point. OpenAI wants people to physically move it through their day the same way they’d carry a phone or a coffee mug.
The design also reportedly includes moving parts. Not a screen, not lights alone (though it has those too, apparently indicating when it’s listening), but actual mechanical motion meant to give the object something close to personality. TechSpot’s reporting quoted the device as being built to have “moving parts that help give it personality,” which is a strange sentence to read about a speaker, honestly. Cult of Mac’s coverage of the same Bloomberg piece described it almost identically: parts that move on their own when the device interacts with a user.
Nobody has said exactly what moves or how. Does it tilt to “look” at whoever is talking? Does a ring segment rotate when it’s thinking through a response? The reporting so far doesn’t say, and that’s probably the single biggest open question about the whole product.
Built With Jony Ive’s Team, Priced Like It
The device is being developed in partnership with LoveFrom, the design studio Jony Ive founded after leaving Apple. OpenAI acquired Ive’s AI hardware startup io for $6.5 billion last year specifically to build this category of product, so LoveFrom’s fingerprints on the design shouldn’t surprise anyone who’s been following the deal.
What that partnership buys you, apparently, is high-quality metal construction and what multiple reports describe as a genuinely premium finish. That’s also where the Apple lawsuit gets uncomfortable. Apple alleges that OpenAI misled one of its longtime manufacturing partners into performing a confidential metal-finishing technique, the kind Apple uses on products like the iPhone, by suggesting the work had Apple’s blessing when it reportedly didn’t. OpenAI has called the claims “careless, aggressive and oddly personal” and says in court filings it is building “something entirely new and different from anything at Apple.” The company has also said it ran an internal investigation after the lawsuit was filed and came away confident that the speaker doesn’t rely on any actual Apple trade secrets.
Apple, for its part, has asked for a preliminary injunction. If a judge grants that, it could delay or complicate the 2027 launch OpenAI is reportedly targeting. That’s still more than a year out, so there’s time for this to get messy in court before a single unit ships.
Sam Altman, in earlier comments about the broader device family, called the screenless approach “simple, beautiful, and playful.” Whether $300 to $400 feels playful to the average buyer is a separate question. That’s a real jump from the $50 to $100 range most people associate with smart speakers, and it puts the device closer to premium audio gear than to a typical Alexa purchase.
It’s also worth noting that $6.5 billion is not a small acquisition for a company still burning cash on model training and data center capacity. Buying an entire hardware startup just to build one speaker would be a strange use of that money, which is why most reporting frames the doughnut as step one of a longer roadmap rather than the whole plan. OpenAI clearly wants a hardware division that can ship multiple products over several years, not a single novelty item that gets discontinued after one generation like so many first-gen gadgets before it.
What It’s Actually Supposed to Do
Strip away the design story and the device is, at its core, a voice assistant. It’s expected to work similarly to ChatGPT’s existing Voice Mode, letting people have back-and-forth spoken conversations rather than issuing single commands and waiting for a canned response. OpenAI reportedly plans to use more advanced models specifically for this device, ones built to remember details about a specific user over time so conversations feel less like talking to a search bar and more like talking to something that already knows you.
The hardware backs that up. Reports describe speaker grills and microphones for capturing and outputting conversation, plus a camera system and additional sensors that let the AI perceive its physical surroundings and respond with actual context. Point it at a messy kitchen counter and ask what to cook, in theory, and it should be able to factor in what it can actually see. Lights indicate when it’s listening, which is a small detail but a meaningful one given how much scrutiny always-on microphones get.
Fonearena’s coverage of the Bloomberg report frames the device as being designed to work in multiple physical positions: in your hand, set on a nightstand, left on a kitchen counter. That flexibility is really the whole pitch. It’s not trying to be a fixed appliance the way a HomePod is. It’s trying to be something closer to a small companion object that follows you around the house and occasionally sits still.
OpenAI reportedly describes the overall device family internally as a “new type of home computer for the AI era.” That’s a big claim for something the size of a hockey puck, but it lines up with what OpenAI has said publicly about wanting AI hardware that isn’t just a phone with a chatbot bolted on.
I’ll admit, when I first read “doughnut-shaped speaker,” I pictured something closer to a toy than a $400 premium device. The mechanical moving-parts detail is the thing that actually changed my mind a bit. That’s a genuinely unusual engineering choice for a product in this price range, and it’s not something companies add just for show since moving parts are expensive to make reliable.
The Apple Problem Isn’t Going Away
It’s worth spending a bit more time on the legal fight, because it’s not a footnote. Apple’s lawsuit against OpenAI centers on claims that OpenAI orchestrated an effort to obtain confidential hardware information tied to Apple’s own future products, and specifically that one of Apple’s manufacturing suppliers was pushed into performing a metal-finishing process associated with Apple hardware. OpenAI’s response has been consistent across multiple filings: the claims are vague, the device is its own thing, and an internal review turned up nothing that violates Apple’s actual trade secrets.
Courts move slowly, and injunction requests don’t always succeed even when a lawsuit itself has merit. Still, the timing is not great for OpenAI. The company is trying to build hype for a genuinely ambitious hardware category, “AI devices that could eventually replace smartphones” is the actual long-term goal reportedly discussed internally, and it’s doing that while fending off a well-funded competitor’s legal team in open court.
There’s also a broader pattern worth noticing here, one that goes beyond this single lawsuit. Ive spent decades at Apple. His new studio is now building hardware for Apple’s most aggressive AI rival, using at least some of the manufacturing relationships and expertise built over that career. Apple clearly sees that as a threat worth fighting over in court rather than just in the market. Whether that’s fair or not probably depends on where you’re sitting, but it’s not hard to understand why Apple’s legal team is nervous.
Why a Speaker First, and Not Something Flashier
It’s a fair question. OpenAI could have led with something more headline-grabbing, glasses, a pin, a phone-shaped device. Instead the first confirmed product is, functionally, a smart speaker. That choice looks less strange once you consider what OpenAI is actually trying to prove.
A speaker is a known category with a known supply chain. Microphones, drivers, batteries, wireless chips, these are all parts that exist and that manufacturers already know how to source at scale. Earbuds like the rumored Sweetpea project need much tighter thermal and battery engineering to survive sitting in someone’s ear for hours. Smart glasses need optics, which is its own extremely hard problem that companies like Meta and Google have spent years on with mixed results. A puck-sized ring you set on a counter or hold in your hand sidesteps a lot of that difficulty.
There’s also a simpler explanation floating around in some of the coverage: OpenAI wants a product that demonstrates its voice and reasoning models in the most direct way possible, without the added complexity of a screen or a wearable form factor competing for attention. No screen means no UI team building menus, no app icons, nothing to tap. Just a conversation. If the pitch is “an AI that behaves like a companion instead of an app,” starting with something that only talks back, and occasionally moves, is arguably the cleanest way to make that case.
What Reviewers and Analysts Are Already Saying
Reaction to the doughnut shape has been, to put it mildly, mixed. Some coverage has leaned into how genuinely odd the concept sounds on paper. TechCrunch’s own headline called the pricing detail “mysterious,” and more than one outlet used the word “bizarre” to describe the ring form factor before getting into the actual specs. That’s not necessarily bad press. Plenty of now-iconic gadgets sounded strange the first time anyone described them out loud. The original iPod’s click wheel, the Nintendo Wii’s motion remote, even the first HomePod’s cylindrical mesh design all got some version of “wait, why does it look like that” coverage before people actually held one.
What’s different here is the price. A $300 to $400 smart speaker is a genuinely tough sell next to a $50 Echo Dot or even a $100 Nest Audio, especially for a first-generation product from a company with zero track record shipping consumer hardware. OpenAI is betting that the combination of Ive’s design pedigree, the premium metal build, and genuinely differentiated AI capability is enough to justify that gap. Analysts covering the story have generally agreed the price only makes sense if the device can do things existing smart speakers simply can’t, deep personalization, real contextual awareness through the camera, and conversation quality that feels closer to talking to a person than issuing commands to an appliance.
A Quick Look at How It Stacks Up
Compared with what’s already sitting on shelves, the differences start to look pretty stark once you line them up. Amazon’s Echo Dot and Google’s Nest Mini both sit in the $25 to $50 range and stay plugged into an outlet permanently. Apple’s HomePod mini runs closer to $99, and the full-size HomePod is roughly $299, putting it right in the same neighborhood as OpenAI’s rumored pricing, minus the battery and the moving parts. Sonos’ higher-end speakers can climb past $400 too, but none of them are trying to double as a portable, camera-equipped AI companion.
So the doughnut isn’t really competing on price with Echo Dots and Nest Minis. It’s competing on category. OpenAI seems to be arguing that “smart speaker” is the wrong bucket entirely, and that this thing belongs somewhere closer to a wearable-adjacent AI device, just one you carry in your hand instead of on your face or wrist.
How This Fits Into OpenAI’s Bigger Hardware Bet
The doughnut speaker isn’t a standalone side project. It’s reportedly the first of a family of AI devices OpenAI plans to release, with earlier reporting pointing to AI earbuds codenamed “Sweetpea” surfacing as early as January. The speaker is expected to be the first product to actually ship, likely because a screenless, largely self-contained speaker is a simpler engineering problem than earbuds that need to handle real-time audio processing on a tiny battery.
A 2027 launch window gives OpenAI roughly a year and a half from now to finalize the design, work through manufacturing at scale, and, presumably, resolve or at least stabilize the Apple litigation. That’s not a lot of time for a company building its first piece of consumer hardware from scratch, even with LoveFrom’s experience shaping the industrial design.
It’s also worth being honest about how much of this is still unconfirmed. Bloomberg’s sources spoke on condition of anonymity because the work is confidential, and OpenAI hasn’t put out an official statement confirming the shape, the price, or the 2027 date. Reports like this from Gurman have a strong track record with Apple-related leaks specifically, so there’s real weight behind it. But “reportedly” is doing a lot of work in every sentence of this story, and that’s worth keeping in mind before anyone starts planning where the doughnut goes on their kitchen counter.
What’s clear is that OpenAI isn’t trying to build a cheaper, chattier version of an existing product category. A battery-powered metal ring with moving parts and a camera, priced above a Sonos One and designed by the person who shaped the original iPhone, is a genuinely strange bet. Whether that strangeness turns into a defining new device category or just an expensive, oddly-shaped footnote in OpenAI’s history probably won’t be clear until units are actually sitting in people’s homes sometime after 2027.