Inside China's smart glasses boom: the four tiers that explain the market

There is no agreed definition of what a pair of smart glasses actually is. People say AI glasses, AR glasses, or smart glasses, and mean four completely different products.

That is a nuisance in the West. In China it is a real problem, because roughly a hundred companies are shipping into the same category at once and the specs alone will not tell you which of them are competing with each other.

Sylvan Shen has a fix for this. She is an XR technical producer at No Ghost, the Emmy-winning immersive studio in London, and she covers China for the Immersive Wire newsletter. She also builds: her team won gold at MIT Reality Hack, and she develops for Meta Ray-Ban and Snap Spectacles. I had her on Metavertising to map the Chinese market, and she opened with a four-tier framework that I have been using ever since.

The four tiers of smart glasses in China

Tier one is audio-first. No display, no camera. You talk to them, they talk back. Sylvan describes these as functionally smart AirPods in an eyewear frame.

Tier two is the portable virtual monitor. You plug them into a laptop or a phone and a screen appears in front of you, useful for private work on a plane or for gaming. XREAL spearheaded this tier, and the better products handle 2D to 3D conversion natively. This category barely exists as a distinct thing in the West.

Tier three is the camera-plus-AI pair. Meta Ray-Ban is the reference design here. A camera captures your day, you share it, and you talk to a model through the frames.

Tier four is AI plus AR. Real display, real spatial awareness, real contextual overlay. Sylvan calls this the meat of spatial computing and the technical breakthrough every Chinese company is chasing.

Everyone wants tier four. Almost nobody is buying it. What sells today sits in tiers one and two, and understanding why is most of the story.

Why the simplest glasses are winning

Sylvan’s read is that the category still lacks rigid demand. There is no task that people cannot do another way, and there is no standardized answer yet to the objection that keeps coming up first.

That objection is privacy. Not the wearer’s privacy, but everyone else’s. It is the top hurdle to all-day wear, and the industry has not solved it so much as routed around it.

Which brings up something worth sitting with.

The copycat assumption, and the brand that inverts it

Western buyers approach Chinese hardware with two suspicions stacked on top of each other. The first is quality, inherited from twenty years of treating China as a copycat nation producing low-quality spin-offs of Western products. The second is privacy, on the theory that a Chinese device is a data risk in a way an American device somehow is not.

Sylvan does not dismiss either one, but she pushes back on the first with numbers. China has spent the last decade moving from low-labor assembly toward high-end engineering, and she cites data putting the manufacturing defect rate down from 25 percent to 5 percent. Chinese firms lead on open-source AI models and on robotics. On the second suspicion, the Chinese companies going overseas are applying for CE certification and building for GDPR compliance, because they have to.

And then there is the part I did not expect.

The pair of smart glasses on the market with the most aggressive privacy position, the one that deleted the camera and the speakers and kept only microphones and a display, is Chinese. Even Realities, founded in Shenzhen by ex-Apple engineer Will Wang, built its entire product around privacy as the first design principle. Palmer Luckey wore the Even G1 to deliver his TED 2025 talk, reading his notes off the lens because TED bans teleprompters but apparently not glasses.

It is Sylvan’s favorite brand in the category, on the grounds of digital minimalism. Go back and reread the two suspicions above. The company that answered the privacy objection most directly is the one from the country the objection was aimed at.

Why Shenzhen moves faster than anywhere else

There is a term for Guangdong province, and specifically for Shenzhen and Dongguan: hardware Silicon Valley. Sylvan cites figures putting over 80 percent of the smart wearable supply chain in that corridor.

The local saying is that you can design a piece of hardware in the morning, assemble it in the afternoon, and mass produce it the next day. Sylvan is quick to say that is an exaggeration. What it accurately describes is density. When the waveguide supplier, the micro LED fab, the frame manufacturer and the assembly line are inside the same hour’s drive, iteration cycles collapse in a way they cannot anywhere else on earth.

If you want to see it rather than read about it, Sylvan points to Cayden Pierce, CEO of Mentra, who filmed his trip through Shenzhen showing how smart glasses actually get assembled.

The hundred-glasses war

Chinese media has a name for what is happening: 百镜大战, the battle of a hundred glasses. Sylvan first saw it in Xinhua.

That level of competition produces one obvious outcome, which is a price war, and one less obvious one. Under that much pressure, and with Meta’s Ray-Ban success sitting there as external proof that the category works, Chinese firms have poured their time into prototyping hardware rather than into brand and user experience. Sylvan thinks that is the deep cause of the perception gap, more than any single marketing failure.

Compare the strategies directly. Meta partners with fashion houses and celebrities, then pushes users to post what the camera captured, so the product’s growth engine is social. Chinese brands mostly do not do this, and Chinese users mostly do not ask them to.

What Chinese users actually buy glasses for

Sylvan describes Chinese consumers as pragmatic. Less weight on how the glasses look and how smooth the interface feels, more weight on whether the thing solves a problem today.

The top use case in China is real-time translation. Not content capture, not social sharing. Translation.

The clearest example of the utility-first approach is Alibaba, which wired its AI glasses into its own commerce stack. There is a feature called look and pay: look at a product to see its price, look at a QR code to pay through Alipay. No fashion collaboration required. The glasses are a limb of an existing super app.

Who is building, and what each of them is really selling

Sylvan splits the field into three groups, and the useful part is that each group is optimizing for something different.

The internet giants, Alibaba and Baidu, are not primarily trying to make money on hardware. Glasses are a distribution channel for their large language models and their existing app ecosystems. Hardware margin is close to irrelevant to that bet.

The phone manufacturers, Huawei and Xiaomi, already have the user base and the mature supply chain. They are extending what your phone already does. Xiaomi’s AI glasses are built into its smart home system, so you use voice or gaze to control the house.

The AR startups, Rokid and XREAL, have neither the user base nor the resources of the other two, so they are trying to win specific verticals outright. XREAL’s niche is the portable monitor, with high-end displays aimed at gaming and productivity, which is exactly tier two from the framework above.

Where China is genuinely behind

Here is the part of the conversation that resists the tidy story.

The Chinese developer ecosystem is Android-focused, while Meta Ray-Ban supports iOS and Android development. RayNeo has done the most to lower the barrier, with an AR development toolkit built on OpenXR, Android and Unity, and there are smart glasses hackathons starting to appear in China. But Sylvan, who develops on both sides, says plainly that the Western developer ecosystem is still ahead.

She frames it as a stage problem rather than a permanent one. Chinese smart glasses are moving out of prototype experimentation and have not yet arrived at mature ecosystem competition. Whether they get there depends partly on whether the hundred-glasses war consolidates first.

The AI question nobody has answered

Multimodal AI is the reason this category woke up. Sylvan’s phrase for glasses is the physical vessel for multimodal AI, and nearly every manufacturer treats its own model as the thing worth showing off.

Except users are not behaving the way the roadmaps assume. Sylvan points to data showing that most Meta Ray-Ban buyers prioritize first-person capture over AI contextual awareness. People bought the camera. The always-listening, always-watching assistant is the part they have not warmed to, and privacy is the reason again.

Her proposal is the most interesting thing she said, and it cuts against every manufacturer’s commercial interest. Give people a choice of model. Let a user run Gemini or ChatGPT or a Chinese open-source model, or switch AI off entirely, with a hardware signal that tells the people nearby which state the glasses are in.

For Chinese brands going global there is an argument for this beyond principle. Their own models are part of what Western buyers are nervous about, so opening the slot removes a barrier. Whether any manufacturer will voluntarily give up owning the assistant layer is a different question, and I do not think anyone knows the answer yet. Meta certainly is not offering it. Neither is Alibaba.

Q&A

What are the four tiers of smart glasses in China? Audio-first glasses with no display or camera; portable virtual monitors that plug into a laptop or phone, led by XREAL; camera and AI glasses in the Meta Ray-Ban mould; and AI plus AR glasses with true spatial computing capability. The framework comes from XR producer and China reporter Sylvan Shen.

What is the “battle of a hundred glasses”? 百镜大战, a term used in Chinese media including Xinhua, describing the roughly one hundred Chinese companies competing in smart glasses simultaneously. It has driven rapid hardware iteration and a price war.

Why are Chinese smart glasses developed so quickly? Supply chain density. Guangdong province, particularly Shenzhen and Dongguan, holds an estimated 80 percent of the smart wearable supply chain, so design, component sourcing and assembly happen within a short radius.

What is the most popular use case for smart glasses in China? Real-time translation, ahead of photo and video capture or social sharing.

Are Chinese smart glasses safe from a privacy standpoint? Chinese manufacturers selling into Western markets pursue international certifications such as CE and comply with GDPR. One Chinese brand, Even Realities, removed the camera and speakers entirely and built its product around privacy as a design principle.

Which Chinese companies make smart glasses? Alibaba and Baidu among internet platforms; Huawei and Xiaomi among phone manufacturers; XREAL, Rokid, RayNeo and Even Realities among specialists.

Is China ahead of the West in smart glasses? On manufacturing speed and supply chain, yes. On developer ecosystem maturity, Sylvan Shen’s assessment is that the West is still ahead, though the gap is narrowing as Chinese platforms like RayNeo open up Unity and OpenXR based toolkits.

Listen to the full episode

The full conversation with Sylvan Shen is on Spotify and Apple Podcasts.

Listen on Spotify

Listen on Apple Podcasts

• 00:00 Intro: who is Sylvan Shen
• 04:08 The four tiers of smart glasses, explained
• 06:22 Why the simplest glasses are winning
• 07:36 Privacy, and the “made in China” question
• 11:35 Hardware Silicon Valley: Shenzhen, Dongguan and the 80 percent
• 14:24 The hundred-glasses war and the branding gap
• 16:00 Alibaba’s look and pay, and what Chinese users want
• 18:22 Internet giants, phone makers and AR startups
• 22:19 The developer ecosystem gap
• 26:32 AI models, freedom of choice, and the privacy signal
• 31:17 Where to find Sylvan

You can follow Sylvan Shen on LinkedIn.

Building for glasses? Apply below for our community of glasses developers.