OpenAI and Jony Ive AI Device Rumours: What We Know So Far

The OpenAI x Jony Ive collaboration is shaping up to be the most-watched hardware debut since the first iPhone.

Let me be upfront from the start: nothing here is for sale yet. As I write this in mid-2026, there is no box to unwrap, no retail listing to scroll through, and no review unit on my desk. What there is, however, is a remarkably consistent stream of leaks pointing toward a family of devices that could genuinely reshape how we interact with AI. OpenAI's Chief Global Affairs Officer Chris Lehane confirmed at Davos that the company will debut its first consumer hardware device in the second half of 2026 — so this is no longer a question of if, but when and what.

Throughout this piece I'll be candid about which claims feel solid and which are still in the realm of educated guesswork. If you're the sort of reader who likes their tech coverage with a dose of healthy scepticism, you're in the right place. Let's dig in.

Concept visualisation for illustration — based on rumoured specs, not a confirmed design.

The Story So Far: How We Got Here

The collaboration between OpenAI and Jony Ive is, on paper, a dream pairing. Ive defined the visual and tactile language of modern consumer electronics across two decades at Apple, and OpenAI has done more than anyone to bring conversational AI into everyday life. The premise behind their joint effort is deceptively simple: if AI is going to be your constant companion, the hardware that delivers it shouldn't look or feel like a smartphone bolted onto a chatbot. It should be something new.

That ambition explains why the early signals point so firmly away from screens. The devices in development reportedly lean into voice, context and ambient awareness rather than the tap-and-scroll paradigm we've lived with since 2007. It's a bold bet, and one that immediately invites comparison to the high-profile AI hardware flops that came before — but more on those later, because the comparison is more flattering to this project than you might assume.

What we've learned has trickled out through a combination of factory leaks, reporting from outlets like The Information, and the odd official soundbite. The headline takeaway is that this isn't a single gadget — it's a roadmap. There are at least three distinct hardware concepts in the pipeline, each with its own internal codename and its own role to play.

The Devices in Development

Let's lay out the lineup as it currently stands. Rather than launching one hero product and hoping it sticks, OpenAI appears to be building a small ecosystem of AI-first devices, staggered across 2026 and 2027.

The star of the show — and the device most likely to land first — is codenamed "Sweetpea," with a commercial name rumoured to be "Dime." It's reportedly a "special audio product" sitting somewhere in the earbud or "open-ear headphones" zone. Think of it less as a pair of AirPods and more as a wearable AI you can talk to, listen through, and carry without ever pulling out a phone.

Sitting alongside it is "Gumdrop," a pen-shaped device that's a little more enigmatic. It's said to be roughly the size of an Apple iPod Shuffle and, like its sibling, is entirely bereft of a dedicated screen. The third concept is a smart speaker pencilled in for early 2027, and beyond that OpenAI is reportedly exploring a smart lamp and even AI glasses. It's an ambitious spread, and it tells you the company isn't thinking about a single product so much as a new category.

Three distinct form factors, one shared philosophy: ambient, screenless, voice-led AI.

Sweetpea: The Headline Wearable

If you only remember one device from this article, make it Sweetpea. This is the one OpenAI is reportedly betting the farm on, and the leaks around it are the most detailed of the bunch.

Physically, the picture that's emerging is rather lovely — which, given Ive's involvement, should surprise nobody. Reports describe two objects, one for each ear, paired with a little egg-shaped, dental-floss-holder-sized charging dock. The whole thing is said to be small enough to wear around the neck or slip into a pocket, on roughly the same scale as an iPod Shuffle. There's even talk of a metal design "reminiscent of an eggstone," which is exactly the kind of poetic, slightly precious phrasing you'd expect to attach to an Ive product.

Under the hood, the rumoured silicon is genuinely interesting. Sweetpea is said to use a 2nm Samsung Exynos chip for some on-device processing, whilst leaning heavily on cloud-based AI for the heavy lifting. That hybrid arrangement makes a lot of sense for a device this small: you want enough local intelligence to handle wake-word detection and low-latency responses, but the truly demanding reasoning happens on OpenAI's servers where there's no battery or thermal ceiling to worry about.

Perhaps the most intriguing detail concerns what Sweetpea is actually for. The custom processor is reportedly designed to let Sweetpea "replace iPhone actions by commanding Siri." Read that again, because it's a fascinating strategic move — rather than trying to rip your phone out of your life entirely, the first device may work through your existing handset, issuing commands and automating tasks via the voice assistant you already have. That's a far more pragmatic on-ramp than the all-or-nothing pitch that sank earlier AI gadgets.

Gumdrop: The Pen-Shaped Wildcard

Where Sweetpea is the crowd-pleaser, Gumdrop is the one that's got me genuinely curious. A pen-form AI device sounds like a gimmick until you read what it's reportedly capable of — at which point it starts to look like the most quietly radical idea in the whole lineup.

Gumdrop is described as roughly iPod Shuffle-sized, pocketable or wearable around the neck, and — like Sweetpea — completely screenless. So far, so familiar. But the standout feature is its contextual awareness: it reportedly packs a suite of sensors including cameras and microphones, giving it an understanding of its surroundings that a pair of earbuds simply can't match.

That ability to run models locally and only escalate to the cloud when necessary is a meaningful design choice. It hints at faster responses, better privacy for routine tasks, and graceful behaviour when your connection drops. The handwriting-to-ChatGPT trick, meanwhile, is the kind of small but delightful feature that could win over a specific crowd — students, journalists, anyone who still thinks better with a pen in hand.

The Smart Speaker and the Wider Ecosystem

The third confirmed concept moves the action from your pocket to your home. OpenAI is reportedly developing a smart speaker for an early 2027 release, with a smart lamp and AI glasses also somewhere on the drawing board.

The speaker is the most "Big Tech" device of the lot, and it comes with a feature set that's equal parts impressive and eyebrow-raising. It reportedly includes an integrated camera and is designed to learn about who is using it and what's around them. There's even talk of a facial recognition feature similar to Face ID, allowing it to identify individual household members.

What sets this apart from the Echos and Nests of the world is the proactivity. Rather than simply answering questions, the speaker is said to observe users and suggest actions to help them reach their goals — the oft-cited example being a nudge toward an early bedtime when you've got a morning meeting on the calendar. That's a genuinely different relationship with a home device: less a voice-activated kitchen timer, more an ambient assistant that watches, learns and intervenes.

I'll be honest — that's also where my enthusiasm meets my caution. An always-on camera with facial recognition that proactively offers life advice is a lot of trust to place in any device, however elegantly designed. How OpenAI handles the privacy framing here will make or break the speaker's reception, and it's the area I'll be scrutinising hardest when it eventually ships.

The $200–$300 target undercuts a premium HomePod whilst sitting comfortably above the budget smart-speaker crowd — a sensible position for a device that's trying to feel special without scaring off mainstream buyers. UK pricing hasn't been mentioned, but a straight conversion would land it somewhere in the region of £160–£240, before the usual tax and import adjustments that tend to nudge US-to-UK numbers upward.

The rumoured smart speaker swaps the touchscreen for cameras, facial recognition and proactive suggestions.

How It Stacks Up Against Rivals

No AI device launches in a vacuum, and OpenAI's hardware will inevitably be measured against the gadgets that tried this first. The two most obvious reference points are the Humane Ai Pin and the Rabbit R1 — both of which arrived with enormous hype and, to put it kindly, mixed results. Here's how the rumoured OpenAI approach compares on the things that actually matter.

The crucial difference jumps out of that table immediately: where Humane tried to convince you to ditch your phone, Sweetpea reportedly wants to work with it. That single strategic choice addresses the number-one criticism levelled at the Ai Pin — that it asked too much, too soon, for too little payoff. Pair that with first-party AI models and Ive-grade industrial design, and OpenAI has at least set itself up to avoid the most obvious traps.

Reading the Signals: A Speculative Scorecard

Since there are no benchmarks to report — no review units exist, and OpenAI hasn't published a single performance figure — I want to be completely transparent about what the bars below represent. This is not a measured test. It's my read on how strong the various signals are around each aspect of the project, based on how consistent and credible the leaks have been. Treat it as an informed confidence gauge, not a lab result.

The pattern here mirrors my overall feeling about the project. The top end — design and AI — is about as strong as a pre-launch story gets, thanks to the calibre of the people involved. The bottom end — privacy, reliability, the nitty-gritty of daily use — is exactly where we have the least visibility, and exactly where these devices will ultimately be judged. That gap is the whole story of AI hardware in 2026.

Manufacturing, Scale and What It Tells Us

Sometimes the most revealing leaks aren't about features at all — they're about logistics. And the manufacturing signals around Sweetpea are loud.

The device is reportedly being built by Foxconn, the same manufacturing giant behind the iPhone, with projected first-year production volumes of 40–50 million units. To put that in perspective: that is not the order book of a cautious experiment. It's the kind of volume you commit to when you genuinely believe you're launching a mainstream consumer hit. Humane and Rabbit shipped a tiny fraction of that. If these numbers are accurate, OpenAI is planning for Sweetpea to be a genuinely high-street product, not a niche curiosity for early adopters.

That September possibility is the one I'd circle on the calendar. It's at the early, optimistic edge of the second-half-of-2026 window, but it would line up neatly with the autumn launch cadence the industry has trained us all to expect. Whether OpenAI hits that or slips toward the back end of the year, the broader takeaway is the same: this is a serious, large-scale hardware effort with the supply chain to match.

The Case For and Against Getting Excited

Let me put my cards on the table with a balanced look at why this project deserves your attention — and why you should keep one eyebrow raised.

Star power and supply-chain muscle are on OpenAI's side — but the category's track record demands caution.

Who Should Be Paying Attention?

Since you can't actually buy any of this yet, "who should buy it" becomes "who should keep a close eye on it." Here's how I'd break it down.

Frequently Asked Questions

Anticipation Rating

Since I can't rate a product I haven't used, this is an "anticipation rating" — my honest read on how promising the project looks from where we stand today, weighing the strength of the people and signals against the genuine unknowns.

The high marks reflect the extraordinary calibre of the team and the cleverness of the phone-companion strategy. The lower marks are honest acknowledgements that we simply don't know how these devices perform, how long they last on a charge, or how OpenAI intends to handle the privacy implications of always-listening, sometimes-watching hardware. Promise is sky-high; proof is non-existent. That's the tension that defines this entire project right now.

Enormous promise, plenty of unanswered questions — the defining tension of OpenAI's hardware push.

I'll be updating my coverage the moment OpenAI moves from rumour to reveal. Until then, take every specification here as the provisional, leak-derived best guess it is — and enjoy the speculation, because there's plenty more of it to come before this thing finally lands on a shelf near you.