Why Are AI Glasses Dropping Cameras Again?

From Addition to Subtraction

By VR Tuoluo ZJ

A few days ago, Apple officially unveiled its new-generation iPhone lineup, with the foldable iPhone Duo drawing the most attention. But to achieve the foldable form factor, the device made trade-offs across many specifications. Its dedicated telephoto camera was cut, it does not support shooting spatial photos, Face ID is gone, and the device weighs in at 254g.

Image source: Apple

Clearly, even Apple struggles to make a product perfect in every way. That is especially true for hardware worn close to the body and designed for extended use. Its previous attempt to "have it all," the Vision Pro, crammed display, computing, sensing, and interaction capabilities into a single device, making the headset extremely heavy and leaving battery life in doubt — hardly suitable for long wear.

The same holds for AI glasses. Looking back over the past few years, ever since Meta put a camera into its AI glasses, the market has seen a flood of camera-equipped products. From photo-taking and video recording to visual recognition and AI Q&A, and then to displays, audio, and more sensors, vendors kept adding features to the frames, hoping the glasses could take on more tasks. Even with efforts to trim weight, the glasses still hovered around 40 grams.

But this year, another trend has taken off. More and more new products are deliberately dropping the camera while keeping the display. That has also prompted people to wonder: when AI glasses truly become all-day wearables, what exactly deserves to stay on your face?

01

No camera + display: another new wave of AI glasses

Back in 2024, Even Realities launched its first AI glasses product, the Even G1. It had no camera and no speaker, but featured binocular monochrome green displays, mainly used to deliver content such as navigation, translation, teleprompting and AI information prompts. At a time when a flood of products were competing to add cameras and more sensors, the Even G1 took a different path and carved out its own route.

Even G1. Image source: Even Realities

This year, that trend has swept through the industry.

Note: Compiled from publicly available information; official sources take precedence.

Judging from products already released or disclosed, vendors position their offerings quite differently. Some emphasize everyday information, some revolve around an AI assistant, and others target meeting and business scenarios directly. Overall, however, these products share one common trait: the camera is removed while the display function is retained.

Even itself continues to iterate along this path. Compared with the first generation, Even G2 sticks to the camera-free, binocular monochrome green display approach, while focusing its main upgrades on display and wearing comfort. Display resolution, refresh rate, field of view and brightness have all improved, and the company has redesigned the optics structure to make the lenses about 30% thinner than the previous generation.

Another product, the RayNeo iO, focuses on AI capabilities and everyday use. It drops the camera and uses binocular Micro-LED displays with diffractive optical waveguides. The whole device weighs about 34 g and is equipped with a four-microphone array and a bone conduction sensor, enabling AI answers, notifications, schedules, live captions, translation and teleprompter content to appear before the user's eyes. Features such as Life Log and meeting notes also rely mainly on voice input.

Image source: RayNeo

Compared with simply plugging in a large language model, the RayNeo iO goes a step further by reorganizing its AI capabilities around all-day wear. Beyond routine Q&A, it aims to give AI long-term memory by continuously capturing voice and contextual information, letting it take part in every aspect of users' lives and work. This camera-free AI glasses product forms a complete interaction loop through voice, memory and display.

Halliday G2 brings this combination into a more clearly defined business setting. Like the others, it has no camera and uses a binocular monochrome green display. Its core feature, Meeting Flow, focuses on voice information in business scenarios, offering users subtitles and translation, decision confirmation, to-do tracking and quick search, while also compiling structured records after a meeting ends.

Image source: Halliday

Even Meta, the flagship vendor that brought camera-equipped AI glasses to the mass market, is now experimenting with camera-free products. The company plans to launch a new pair of smart glasses this fall, internally codenamed Luna. The biggest change is the removal of the camera, while retaining microphones, speakers and AI capabilities. It has not yet been confirmed whether Luna will feature a display, but the move at least shows that the camera is shifting from a default feature of AI glasses to a capability that can be included or dropped based on a product's positioning.

In addition, VR Gyro has learned from sources that two leading domestic AR glasses makers are also preparing AI glasses products of this kind, with launches possible as early as the fourth quarter of this year or around CES.

Overall, these smart glasses target different needs, but the trade-offs behind them are now clear. AI glasses are no longer trying to handle every task at once — capture, display, recognition and AI — but instead concentrate their functions on specific scenarios where they need a display and continuous access to voice information, though not necessarily a camera.

As similar hardware combinations begin appearing in more and more products, it also reflects that manufacturers are rethinking the value of placing every hardware component and feature on AI glasses.

02

Cut the camera, keep the display: makers are recalculating the numbers

So why is the camera the first thing to go?

The burden of a camera goes far beyond adding a sensor to the frame. To handle both the visual processing the camera demands and all-day battery life, many AI glasses use a heterogeneous dual-chip design: a high-performance chip handles the camera, image processing and heavier AI computation, while a low-power chip takes on voice, audio, Bluetooth connectivity and other tasks that must run for long periods, so the high-performance chip does not have to stay active all the time.

Image source: Meta

Remove the camera and things get considerably simpler. Without continuous image capture and visual computing demands, some products no longer need a separate high-performance vision processing platform for the camera; a single low-power SoC can handle audio, connectivity, display and basic AI tasks. RayNeo iO is one example, using a single BES2800 from Bestechnic. Compared with a dual-chip design, this architecture cuts PCB footprint and system complexity while freeing up room to lower overall power consumption and BOM cost. With fewer chips, the glasses gain more leeway to optimize both price and battery life.

This system-level simplification ultimately shows up in the wearing experience, with weight being the most obvious change. Take Ray-Ban Meta as an example: the standard model weighs about 48 g, and other camera-equipped AI glasses commonly seen on the market also largely stay in the 40 g range. This year, however, the picture has changed.

With the cameras removed, this batch of display-equipped AI glasses has been trimmed further to around 30 g. RayNeo iO weighs about 33 g, Even G2 around 36 g, and the next-generation display AI glasses announced by Kanan Technology could go as low as roughly 26 g, very close to the weight of ordinary eyeglasses.

At the same time, advances on the display side are supporting this form of AI glasses. As the technology continues to break through, display systems that once took up a large amount of frame space are becoming smaller, lighter and better-performing. Judging from the products showcased by upstream exhibitors at this year's CIOE, Micro-LED-based microdisplay modules continue to shrink in volume, with some monochrome optical engines now fitting within 0.1 cc or even 0.05 cc, further reducing weight while maintaining display quality and mass-producibility.

Image source: VR Tuoluo

Waveguides are also getting thinner and lighter. At CIOE, Goertek Optics presented several diffractive waveguide designs for lightweight AI/AR glasses. Its F20Gi, made using nanoimprint lithography, is just 0.5mm thick and weighs 3.2g. Another fully laminated, etched full-color diffractive waveguide, the F20Ge-FL, measures only 0.65mm in total thickness including two cover layers and weighs under 4g. At the same time, advances in DUV lithography, etting and nanoimprint processes are pushing diffractive waveguides toward a better balance of low weight, display quality and mass manufacturability.

These upstream technology changes let manufacturers deliver decent display quality even after removing the camera, while pushing the whole pair of glasses toward a lighter form factor closer to ordinary eyewear.

Beyond hardware, cameras introduce a problem that hardware iteration alone cannot easily solve: privacy. In business meetings, offices and other everyday settings that require long hours of wear, a camera permanently facing outward can easily make nearby people uneasy. The wearer knows whether they are recording, but bystanders have no quick way to tell. The privacy controversies Meta has recently found itself in show that as these devices become more widespread, concerns like these draw ever more attention.

Image source: Meta

Removing the camera at least sidesteps the most immediate bystander anxiety about being filmed, making such products easier to deploy in scenarios that emphasize extended wear.

In terms of information-gathering efficiency, numerous studies suggest that roughly 70-80% of the information people take in comes through vision. In AI glasses, that advantage is even more direct. Whether for navigation or information prompts, displaying content right before the eyes lets users glance up and grasp key information in an instant. That is precisely why makers are still willing to reserve structural, power and cost headroom for a display after dropping the camera.

Cameras still have value, but for products that emphasize light weight, privacy and long-duration wear, they are no longer a must-have feature.

03

From Adding Features to Making Choices: AI Glasses Are Diverging

But the growing number of camera-free products does not mean the camera-based approach will be phased out. For users who want to use glasses as a first-person recording device, or who rely on visual recognition and multimodal AI, cameras remain indispensable.

The more likely change in the AI glasses market, therefore, is a further split in product approaches.

One group of products continues to include cameras and audio, making capture and recording, visual understanding, and AI its core capabilities. These products need no display, have a relatively simple structure, and have already demonstrated consumer market demand through existing products.

Another group drops the camera to prioritize weight, power consumption, and space for the display, aiming for longer daily wear through features such as subtitles, navigation, translation, notifications, and AI information prompts.

A third group retains cameras, a display, and full AI capabilities, handling both visual input and information output in hopes of covering more interaction scenarios. Compared with the first two, this approach has a higher functional ceiling and comes closer to manufacturers' vision of next-generation spatial computing terminals. For example, RayNeo X3 Pro and INMO Air 3 are deeply invested in this approach, and Rokid's next-generation AI glasses, set to debut on September 24, will continue along the same path.

None of these approaches is inherently superior, because they address different needs. For AI glasses, this is an inevitable sign of a market that is gradually maturing.

While the market was still in its early stages, manufacturers kept adding features to demonstrate what a pair of glasses could do. As products genuinely began to be used by consumers in daily life, their functions were naturally reselected, and people with different needs started deciding which features were better suited to stay on their own faces. This reflects the trends of "multiple glasses per person" and "a different pair of glasses for everyone."

       Original title : AI Glasses: Why Are the Cameras Being Dropped Again?