The long-anticipated launch of the Apple Vision Pro came as a real disappointment to the immersive tech industry. After the industry hoped it would be the iPhone moment, it failed to live up to expectations.
IDC put Apple’s headset at roughly 45,000 units in the final quarter of 2025, against about 390,000 across its 2024 launch year. This led Apple to cease production and cut more than 95% of its ad spend (Hypebeast, Jan 26).
As the headset market stalled, the immersive tech industry’s hopes turned to smart glasses, and that bet paid off. EssilorLuxottica reported that its Meta AI glasses sales more than tripled in 2025 (CNBC, February 26), and by the first quarter of 2026 the category reached 2.25 million units in a single quarter, up 167% on the year before, with Meta taking 69.2% of the market (IDC, 26).
The industry didn’t fully anticipate the mass public backlash over privacy and the misuse that came with it. That misuse earned the glasses a nickname that stuck, pervert glasses.
The bet that failed and the one that paid
The Vision Pro failed on several grounds the industry had set for itself. At $3,499, sold in only 13 countries, heavy on the face and thin on native software, it stayed where every high-end headset has stayed, with early adopters (PYMNTS, 26). Morgan Stanley put the failure plainly, naming the cost, the form factor and the lack of native apps as the reasons it never sold broadly. None of that was a surprise. Every one of those problems was clearly visible at launch, and the industry bought the promise anyway.
The Ray-Ban Meta glasses won by ignoring the promise. They run no spatial operating system and do something narrower, taking photos, capturing video, playing audio, answering questions through Meta AI and translating speech, all from a frame that reads as ordinary eyewear and costs a fraction of a headset. That partnership with EssilorLuxottica put Meta at the front of the smart glasses market (IDC, 26). Buyers didn’t want to wear a computer. They wanted a camera and an assistant nobody else would notice, and that is the whole problem.
The assistant is the actual product
People are paying for the assistant. Meta AI in the glasses answers questions about whatever you are looking at, translates a conversation or a street sign as it happens, reads your messages aloud and captures a moment without you reaching for a phone. For people with reduced sight or hearing it becomes an accessibility device, live captions of whoever is speaking, or a sighted volunteer guiding them through a room by seeing what the camera sees. The newest version adds a wristband that reads the muscle signals in your hand, so the assistant runs without a screen or a spoken word. That kind of usefulness is what the headset never had, and it is what turned a novelty into something people wear all day.
That value has a direction, and it runs the same way the privacy problem does. An assistant is only as useful as what it can perceive, so the road ahead is more perception, not less. Meta is already building an assistant that watches and listens continuously and prompts you before you ask. The more the glasses understand your world, the more they take in of everyone else’s. The same perception that translates the stranger in front of you can record them without their knowing. Value and harm are one capability seen from two sides, which is what makes pervert glasses a design problem rather than a communications problem.
Why the pervert glasses backlash is a market problem
The privacy backlash matters more than a normal product controversy because it lands on the one device the industry is winning with, not the one it already lost. In 2025 and 2026, the pattern repeated in public. The University of San Francisco warned that a man was approaching women on campus wearing the glasses and recording them. A woman found she had been filmed by a stranger only when the clip reached people she knew (CBS News, 26). By late July, Instagram was removing harassment videos shot on the glasses and had deactivated two creator accounts with more than a million followers each (Fortune, July 26).
Institutional responses moved faster than the industry expected. A UK events company banned wearable recording devices from its Comic-Con dates after guests raised concerns. US lawmakers pressed Meta over facial recognition plans for the glasses (Yahoo, 26). The Electronic Frontier Foundation told people to think twice before buying or using them (EFF, March 26). None of this stops the glasses from selling this year, but it does cap how far they can go. A device only reaches smartphone scale once wearing it in a cafe, a gym or a school is unremarkable, and every viral clip of a woman filmed without consent pushes that permission the other way. Growth and trust have become the same problem.
Where the iPhone comparison breaks
The iPhone comparison breaks over consent. People point to the sharp jump in 2025 sales when they call this an iPhone moment (247 Wall St, May 26). It flatters the hardware and skips the hard part.
The iPhone changed one person’s relationship with their own device. Smart glasses change a stranger’s relationship with everyone around them. When someone lifts a phone, you can see them do it, but glasses point at you the whole time and never signal when they are recording, and that is a different social contract no volume curve settles. The real inflection for immersive tech is not the day the glasses get cheap enough. It is the day the person across from you stops assuming you are filming them, and no version of that day arrives on price alone.
Meta’s real advantage is the part the iPhone story ignores, which is distribution. EssilorLuxottica owns Ray-Ban, Oakley and the retail and prescription-lens network behind them. That is why the glasses look normal and reach shelves no pure tech company can access, and why a trust failure is expensive. When harm attaches to a fashion brand people already wear, it doesn’t stay contained to a gadget review.
What comes next for immersive tech
Smart glasses become the mainstream device from here, and the headset never will. The next phase is already dated. Samsung and Google launch their first Android XR glasses in autumn 2026, built with Gentle Monster and Warby Parker frames on Qualcomm silicon, with display versions following in 2027 (9to5Google, May 26). Apple, having pulled back on the Vision Pro, is reportedly redirecting towards its own glasses. The headset doesn’t disappear, but it settles into what it always was, a niche for work, training and high-end media rather than a mass device. The centre of the market is a lightweight camera-and-AI frame, and by the end of 2026, three of the largest technology companies are selling one.
That turns the contest away from specs. When every major frame captures video and runs an assistant, hardware stops deciding the winner. Three things decide it instead. Whether the assistant earns its keep is one, useful enough that people accept the camera as the price of it, which is why Meta is testing more capable super-sensing AI even as the privacy questions grow (Fortune, July 26). Trust design is another, built so a bystander can read it and not only the wearer. Meta already fits a recording light that cannot be switched off and now cuts the camera if the light is tampered with. The whole category has to move that way, towards signals aimed at the people being filmed rather than the people holding the frame. Regulation is the last of the three, and it arrives whether the industry shapes it or not.
The advantage goes to the company that builds consent into the product instead of bolting it on for compliance, because consent is what widens the group willing to wear one. Whether smart glasses become the next computing platform or stay a clever accessory will not be settled by the next chip. It comes down to whether the person sitting opposite you is comfortable with you wearing it.
