The market is bullish on AI glasses, but this founder says the next two years will be harder
He reckons the current smart-glasses market is only two to three million units, facing squeeze from Apple, Meta, and domestic tech giants—the real inflection point waits for supply chain, content ecosystem, and product to mature together in 2028–2029.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
The faster phones connect us, the more isolated we become
Yang sees phones as solving connection efficiency—you can reach a friend across the world in seconds. But phones created information silos: the faster we connect, the more distant we become from those around us. Many people who chat online fall out of touch. Glasses could bridge this gap. Walking past a stranger, both of you see each other's information in real time, spot shared interests and mutual friends instantly, breaking the social island effect. This was his core principle when designing the glasses product.
— Longsheng YangSocial is the killer app for smart glasses
Yang is convinced that social—not translation or photography—is the endgame killer app. His team built a small dating-scene demo: a dozen people wearing glasses at a speed-dating event. The glasses display identity information and the AI guides the conversation, replacing the host or matchmaker role, removing awkwardness between strangers. He frames it as a tiny sample, but one that will scale into daily life and play an ever-larger role in social interaction.
— Longsheng YangThey could read microexpressions—and chose not to
When asked whether emotion recognition could detect if someone is lying or assist ‘cold reading,’ Yang directly answered: ‘Technically feasible, but we don't recommend it.’ Co-host Qin Kuang added that over-datafying someone's emotions is risky: ‘If you data-fy everything, it's genuinely scary.’ Yang explained the company's core principle is tech for good: information should assist people, never replace human judgment or reshape social boundaries. Even if microexpression recognition is ready, a single false read shouldn't destroy a relationship—‘You start as a matchmaker and end as a home-wrecker.’
— Longsheng Yang / Kuang QinHe picked me: the moment I saw Oculus
Yang was struck by “Black Mirror” in sophomore and junior year, frantically searching for similar products. As graduation neared, Oculus's first headset shipped. When asked if he felt chosen by fate, he said: ‘He picked me—it's just that he told me there seemed to be a path to get close to what I wanted to do.’ He immediately flew from Dalian to Shenzhen, found Chaihuo (柴火) Makerspace, the only place to touch VR then. After his resume was rejected, he showed up in person, happened to meet a manager, and got hired on the spot.
— Longsheng YangHeard the hype cycle five times, still believes
In 2016–17, Yang concluded VR was a home entertainment device, too heavy for mobile carry. He pivoted to AR, seeing it as capable of a larger market and becoming the next-generation terminal. Asked whether ‘year one’ hype—heard at least five times—ever shook his faith, he said: ‘I definitely still believe it, and I believe it even now, because the core need itself hasn't changed.’ The need remains constant: floating screens from sci-fi, social tags visible as you walk down the street. He has never told the press this is the year for glasses.
— Longsheng YangBet on all-in-one five years ago, never touched modular design
From the first generation, Yang committed to integrated, lightweight design, sacrificing resolution to get there. Many competitors used DLP, LCOS, or external power packs (split-body systems). Inmo ‘never went that route.’ They're among the rare early companies to never accept external power from day one—Yang calls them ‘the all-in-one pioneer.’ That bet was five years ahead of the industry; only in the last year or two have monochrome glasses started clustering around lightweight design.
— Longsheng Yang / Kuang QinCould record a week of life, chose local storage instead
Inmo Future has 18GB of storage, enough to record 7–10 days of content. For privacy reasons, this capability is stored entirely locally and accessed by the consumer alone—never uploaded to the cloud. On whether to build a feature, Yang applies three gates: first, is there a cohort of people who absolutely must buy it? Second, can current tech and supply chain support it, and does cost match what consumers perceive as value? Third, is the return on investment worth the effort? There's also a values line: even if it makes money, he won't touch projects that circle money through hype, like some web3 ventures—‘we'd walk away from any amount of money.’
— Longsheng YangThree curves converge simultaneously in 2028–2029
Yang identifies two major obstacles to mainstream adoption. First, upstream supply chain—chips, optics, batteries all need a meaningful next step. He predicts three years to support a more refined ‘iPhone 4 moment’ product. Second, content ecosystem: major platforms like ByteDance and Alibaba only began laying out agent strategies around glasses this year; he estimates two to three more years to grow a real one. When hardware, supply chain, and content ecosystem mature and intersect at the same moment, the market will experience a shift fundamentally different from phones, compelling real consumers to buy in.
— Longsheng YangIn their own words · checked verbatim
‘I've always believed social is definitively the killer app at the endgame.’
其实我一直会坚信说 社交它一定是那种最终局的 杀手级应用
Longsheng Yang9:02
‘He picked me—or to put it differently, he showed me there seemed to be a path where I could get close to what I wanted to do.’
他被我选中了 就只能说是 他告诉了我 好像有一条路 是可以能接近我想做的事
Longsheng Yang17:07
‘I definitely still believe it, and I believe it even now, because the core need itself hasn't changed.’
肯定还是相信的 包括直到现在还是相信的 因为它这个需求本质是不变的
Longsheng Yang21:09
‘We're the pioneers of all-in-one, I'd say—we've always had our own pursuit in design, in lightweighting, in aesthetic sense.’
我们算一体式的鼻祖吧 就是在设计上 在这种清凉化上 时尚度上面 一直是有自己的一些追求
Longsheng Yang36:17
‘In the past you had to search through different interfaces and filter information. In the future, how information flows will definitely be active push, definitely fast retrieval, triggerable with a single word.’
以前我们是需要去通过不同界面的搜索 然后去筛选信息 未来整个信息的传递方式 它一定是这个主动推送的 一定是快速获取的 一句话就可以触打的
Longsheng Yang58:31
Figures
| Current implementation of Black Mirror scenario | 30–40% | 5:02 |
| Air 1 single video plays on Bilibili | Over thirty million | 26:12 |
| Inmo Future storage capacity | 18GB; records 7–10 days of content | 40:18 |
| Projected AI-glasses market size (next 2–3 years) | Two to three million units | 52:28 |
Glossary
- Split-body design / modular system
- A smart-glasses design where the frame connects via tether to an external battery or compute pack for power and processing.
- Product-market fit
- The point when a product achieves real market demand at scale and proves its core proposition to the market.
- SLAM (simultaneous localization and mapping)
- A technique where a device builds real-time spatial maps and locates itself using onboard sensors, without external infrastructure.
- Inside-out tracking
- Spatial tracking using only the device's onboard sensors, requiring no external base stations or markers.
How to listen
Practitioners and investors tracking AI glasses and AR hardware startups, plus anyone curious why social might be the killer app for the next-generation terminal.
The opening discussion of live-streaming tipping history and small talk (0:00–1:16) has low information density—skip straight to the Black Mirror topic.