Meta's AI Pendant Is the Wrong Answer to a Question Nobody Asked

Meta's Reality Labs division burned through $4.2 billion in the first quarter of 2025 alone, a staggering figure even by the standards of a company that has now poured more than $80 billion into its hardware ambitions since 2020. The unit's VR headset sales were declining. Its much-hyped metaverse remained a punchline. And yet, according to internal documents obtained by The Information and reported by Reuters on May 29, Meta is not retreating. It is pivoting hard, away from virtual reality and toward wearables that sit on your face or hang around your neck. The target: 10 million units sold in the second half of 2026. The plan includes an always-listening AI pendant that will begin testing within the next year, plus up to four new smart glasses models. Reality Labs is betting its future on the belief that the smartphone's reign is ending and that AI hardware worn on the body is the next platform.

This pendant will fail, and it deserves to. The AI hardware category has spent two years proving that miniaturising a voice assistant into a clip-on accessory solves a problem nobody walking around in daily life actually has. The smartphone already listens, already summarises, already transcribes, and it does all of it with a screen large enough to correct the mistakes. A pendant that does the same thing without a display is not a revolution. It is a downgrade with a magnetic clasp. Meta's glasses, by contrast, have earned a tentative case for existing. This review of the landscape, built from leaked roadmaps, public financials, and the smoking craters left by previous entrants, is an argument that the industry is backing the wrong form factor and that only one wearable AI device has a plausible claim on the next eighteen months of your attention.

The leaked roadmap, first detailed by Android Headlines citing The Information's reporting, reveals a company racing to diversify its hardware portfolio before Reality Labs's losses become politically unsustainable inside Meta. Four new smart glasses models are in development, spanning different price points and use cases. The pendant, described as an audio-focused device capable of passively capturing and processing conversation throughout the day, is the wildcard. TechCrunch reported that the pendant builds on Meta's earlier acquisition of Limitless, a startup that made a wearable meeting transcription device. The logic is straightforward: if people will wear a camera on their face, they might wear a microphone on their chest. The flaw in that logic is the subject of the next several years of product reviews.

To understand why Meta's pendant will fail, start with the Humane AI Pin. Launched in 2024 to a crescendo of venture-backed hype, the $699 pin promised to free users from their smartphone screens. It projected a laser display onto the palm, responded to voice commands, and took photos. Reviewers at nearly every major outlet panned it. The device overheated. The projection was illegible in daylight. The AI responses were too slow to justify not pulling out a phone. By February 2025, TechCrunch reported that HP had acquired most of Humane's assets for $116 million and immediately discontinued sales of the AI Pin. The device was dead in under a year. HP did not buy the pin business. It bought the talent and the intellectual property, repurposed into an enterprise laptop chatbot platform called HP IQ, as Gizmodo later detailed. The most hyped AI wearable of 2024 ended its life not on anyone's lapel but as a piece of workplace software on a business laptop.

The Humane story is not an outlier. It is the template. The Rabbit R1, a $200 bright-orange handheld AI device that launched in early 2024, rode a wave of CES buzz into a brutal reception. Reviewers called it slow, buggy, and redundant next to any modern smartphone. The Verge reported in January 2026 that Rabbit had announced a completely new device, a 'cyberdeck' for vibe coding, while pushing a major software update to the R1 in an attempt to salvage the original hardware. A $200 AI gadget that launched as a smartphone replacement is now being repositioned as a companion device while its maker pivots to a different category entirely. The pattern is consistent: launch a standalone AI device, discover the phone does it better, pivot or die.

Which brings the analysis back to the pendant. Meta's device, as described across multiple reports citing the leaked internal memo, is designed to listen continuously, transcribe conversations, and surface AI-generated summaries and action items. It is, in essence, a wearable meeting recorder that never stops recording. The privacy implications are so obvious they barely need enumerating, but TechRepublic noted that the pendant raises 'new privacy questions for workplaces and social settings' and that EMEA regulatory friction is likely. A device that records every conversation within earshot, processed through Meta's AI infrastructure, will face the General Data Protection Regulation in Europe and an increasingly privacy-hostile regulatory climate in the United States. The regulatory burden alone makes mass adoption improbable before a single unit ships.

But the deeper problem is not regulatory. It is that the pendant's core use case, capturing and summarising conversation, is already handled by software running on phones and laptops. Apple Intelligence, Google's Gemini, Microsoft Copilot, and a dozen third-party apps already offer real-time transcription and summarisation. They do it using microphones people already carry. The pendant adds a dedicated microphone array and removes the screen. That is a tradeoff with no upside for the user, only for Meta, which gains a new data stream. A wearable that collects continuous audio in exchange for capabilities already available elsewhere is not a product. It is a surveillance device with a value proposition problem.

The glasses are different, and the distinction matters. Meta's Ray-Ban smart glasses, now in their second generation, have sold respectably. The Verge noted in May 2026 that both generations were on sale, a sign that inventory is moving and Meta is managing the product lifecycle like a mature consumer electronics company rather than a science project. The glasses do a few things well: hands-free photo and video capture, phone calls, music playback, and AI queries triggered by voice. They look like Ray-Bans. That last part is the entire strategy. A wearable that passes as a normal accessory has a chance. A wearable that announces itself as a piece of technology does not. The Ray-Ban Meta glasses have found a narrow but real market because they solved the social acceptability problem first.

The numbers bear this out. Reality Labs' $4.2 billion quarterly loss, reported by Gadget Review via Yahoo Finance, is unsustainable in isolation, but the glasses are the one bright spot in a division that has otherwise been a furnace for shareholder capital. Meta's pivot to wearables, including a reported 'Wearables for Work' enterprise subscription, signals that the company sees the glasses as the beachhead and the pendant as a flanking maneuver. The glasses have product-market fit in the way that fitness trackers had product-market fit in 2014: narrow, enthusiast-driven, but real and growing. The pendant has no equivalent anchor.

The competitive landscape reinforces this asymmetry. The Los Angeles Times reported in May 2026 that Google and Samsung had unveiled designs for co-developed AI smart glasses. CNBC reported that China's Rokid is selling AI display glasses with virtual screens that are outselling cheaper models, driven partly by demand for teleprompting features. Snap has launched its $2,195 Specs AR glasses, per The Next Web, with a 51-degree field of view, four-hour battery life, and AI from both OpenAI and Google. The stock dropped 5% on the announcement. Everyone is betting on glasses. Nobody credible is betting on a standalone AI pendant, except Meta, and Meta's bet appears to be an extension of the Limitless acqui-hire rather than an organic conviction that the form factor is correct.

Apple, meanwhile, is watching. MacRumors reported in May 2026 that Apple is experimenting with an AI wearable described as a pin or pendant, potentially launching as soon as next year, alongside AI smart glasses and camera-equipped AirPods. Apple's entry into a category typically signals that the category has been de-risked by someone else's failure. The company rarely ships first. It ships when the component costs, manufacturing processes, and user expectations have been calibrated by a competitor's expensive mistakes. If Apple is prototyping a pin, it is because Humane and Meta have demonstrated what not to do. That does not mean an Apple pin would succeed. It means Cupertino sees enough signal in the noise to warrant the research budget.

Here is the question every AI hardware product must answer, and almost none does: who is this for, exactly, and is that group large enough to ship to? The Humane AI Pin was for a hypothetical smartphone-eschewing user who did not exist in meaningful numbers. The Rabbit R1 was for early adopters who wanted a dedicated AI gadget, a group that turned out to be a few tens of thousands of people, not millions. Meta's glasses are for people who already wear glasses or sunglasses and want basic hands-free utility. That group is large and legible. Meta's pendant appears to be for people who attend enough meetings to want a dedicated transcription device but who also want that device listening during their commute, their lunch, and their conversations with friends. That group is tiny, and the social friction of wearing an always-recording pendant will keep it tiny.

The enterprise angle, the 'Wearables for Work' subscription reported by The Next Web, is a rational attempt to find a buyer with a compliance budget rather than a consumer with a privacy instinct. An employer-mandated AI pendant that records meetings and generates summaries is a product that procurement departments might actually purchase. It solves the adoption problem through mandate rather than desire. But it also caps the addressable market at the size of the enterprise wearables budget, which is orders of magnitude smaller than the consumer market Meta needs to justify Reality Labs's burn rate. An enterprise pendant is a niche product for a division that needs to ship at iPhone scale.

Eighteen months from now, the AI hardware landscape will look like this. The Humane AI Pin will be a footnote in business school case studies about product-market fit. The Rabbit R1 will be an open-source curiosity kept alive by a small community of tinkerers. Snap's Specs will either have found a developer ecosystem or joined the Google Glass graveyard. Meta's glasses, now on their third or fourth generation, will be a modest success, selling in the low millions and justifying their existence inside Ray-Ban's product line. And the pendant will be in one of two places. Either it will have been killed quietly, its team folded into the glasses division, or it will be shipping in small volumes to enterprise customers while consumer adoption languishes in four figures. There is no scenario in which the pendant becomes a mass-market device. The form factor is wrong, the privacy calculus is worse, and the smartphone already won.

This is not an argument against AI hardware. It is an argument against AI hardware that ignores the hierarchy of socially acceptable wearables. Glasses sit at the top. They have a century of cultural permission to exist on the human face. Watches and fitness bands sit on the wrist, another socially settled location. Earbuds occupy the ear. A pendant around the neck, recording everything, is a new category with no cultural permission and no demonstrated demand. Meta can spend billions trying to create that demand. The Humane AI Pin proved that billions are not enough when the product answers a question nobody asked. The glasses have a shot. The pendant is a tombstone waiting for a date.

The one variable that could change the calculus is Apple. If Apple ships an AI pin that integrates seamlessly with the iPhone, leverages the U1 chip for spatial awareness, and gets the industrial design right, it could create a market where none exists. Apple has done this before: the Watch was not the first smartwatch, but it was the first one that sold in the tens of millions. The counterargument is that the Watch succeeded because the wrist was already a proven location for wearable technology and because Apple gave it a clear, singular use case: fitness and notifications. An AI pin has no equivalent anchor use case. Apple's track record suggests it will not ship until it finds one. That it has not shipped yet is itself a data point.

The AI wearable market will grow. Forbes reported in early June 2026 that AI is driving wearable tech toward becoming the next platform for personal computing, with smart glasses and AR headsets positioned as the long-term smartphone replacement. That thesis is defensible. The transition from pocket to face will take a decade, not a year, and it will be led by devices that look normal, weigh little, and do a few things reliably well. The devices that try to jump the line, that skip the social contract and demand new behaviours, that substitute a clip for a screen, will fail. Meta's pendant is one of those devices. It will be reviewed, it will be scrutinised, and then it will be forgotten. The glasses will still be on shelves. The phone will still be in pockets. The AI hardware revolution will arrive, but it will not arrive around your neck.