Is the era of AI glasses here?Article author and source: 36Kr
Fourteen years ago, a live-streamed skydive from a conference venue introduced Google’s AR glasses concept to the public for the first time, convincing Zhu Mingming, who was then deeply focused on cutting-edge technologies, that AR glasses would become the next optimal platform for human-computer interaction, following PCs and smartphones.
Two years later, he left Alibaba to found Rokid, diving headfirst into the untamed AR industry.
During the era when waveguide technology was immature, device battery life was limited, and consumer use cases remained absent, AR quickly shifted from a hot investment trend to an overlooked niche, causing many startups to exit the market.
Rokid experimented across smart speakers, industrial AR, and cultural tourism solutions, quietly waiting for the industry’s turning point. Until the surge of general-purpose large models provided AR glasses with long-missing practical use cases—triggering a surge in demand for interactions like translation, meeting transcription, and real-world object recognition—reviving the once-sleeping sector.
Rokid is actively advancing its plans to list in Hong Kong, with an internal target listing window of July this year. Prior to this, Rokid has completed several capitalization preparations, including completing its share restructuring and attracting multiple industry chain companies to participate in equity placement, covering areas such as optics, chips, batteries, and communications.
Behind the glamour, competition in the industry is intensifying. On one hand, AR glasses manufacturers continue to struggle with massive losses; on the other, domestic tech giants in mobile and internet sectors are rapidly entering the market thanks to their mature ecosystems, while overseas giant Meta has already monopolized the global market share—ushering in a fierce battle among contenders.
Having survived a decade-long winter and secured its entry ticket, Rokid now faces questions from capital markets about profitability while striving to maintain its foothold amid encirclement by industry giants.
I. A company that waited ten years
In 2012, Google co-founder Sergey Brin appeared onstage wearing Google Glass, while engineers jumped from the top of a building in San Francisco, streaming live footage via Google Glass to the event venue, sparking global attention toward wearable smart devices for the first time.
Among them was Zhu Mingming, then head of Alibaba’s M Lab, who was leading research in deep learning and computer vision. Years later, he repeatedly expressed the same assessment on multiple occasions: AR will become the next generation of interactive platforms after smartphones, and AR glasses will be the optimal product form for human-computer interaction following PCs and smartphones.
Two years later, Zhu Mingming left Alibaba to start Rokid, betting on AR.
The industry atmosphere was highly optimistic at the time. In April 2014, Google Glass became available for public purchase, and months later, Microsoft unveiled the HoloLens—a headset touted as the next-generation computing platform after PCs and smartphones—pushing optimism to its peak. Many believed that smartphones would eventually be replaced by AR devices.
But the good times didn't last.
In early 2015, Google officially announced the withdrawal of Google Glass from the market; in the following years, Microsoft’s HoloLens story became clearer, targeting enterprise and professional use cases; Magic Leap, the U.S. AR company that was once wildly追捧 in the primary market and reached a valuation of billions of dollars, failed to deliver on its promises in subsequent product releases, and its valuation narrative gradually collapsed.
The entire AR industry fell into a prolonged downturn lasting several years.
The reasons are not complicated. On one hand, core technologies such as display modules, waveguides, chip performance, and battery life have yet to mature. A truly functional AR headset with display capabilities often means a heavier device, higher costs, and a poorer wearing experience.
On the other hand, even if technical issues are resolved, the industry has still not identified sufficiently strong use cases. For most users, AR glasses priced at thousands or even tens of thousands of dollars cannot replace smartphones.
Looks cool, but has never become a necessity. This has caused capital markets to remain cautious about AR companies over the long term. Over the past decade, numerous AR startups have vanished during industry downturns.
Rokid became one of the few players to persist.
From smart speakers to AR glasses, from consumer markets to industrial applications, from cultural and tourism displays to enterprise solutions, the company has continuously adjusted its direction to secure room for survival.
The turning point occurred after 2023.
With breakthroughs in large model technology, AI is endowing glasses with new value.
Previously, AR glasses addressed the question of what you could see, while AI glasses solve the problem of understanding the world in front of you and providing immediate responses. Capabilities such as real-time translation, meeting transcription, visual recognition, and intelligent Q&A are now giving glasses genuine practical applications.
The AR glasses industry is regaining capital attention.
Meta was the first to validate the market, with Ray-Ban Meta sales continuing to grow; Google has re-entered the smart glasses arena; and Chinese tech companies such as Alibaba, Baidu, and Xiaomi are gradually joining, betting on AI glasses.
After more than a decade of industry cycles, a turning point has finally emerged, and Rokid has reached its moment of capitalization.
In March 2026, the company completed its transformation into a joint-stock corporation, removing institutional barriers to future capital market listing. In the following months, Rokid successively brought in six industry chain enterprises—Huace Film & Television, Kangnai Optical, Haopeng Technology, AWIC Electronics, Quectel, and Jingzhida—achieving comprehensive coverage across core segments of the industry chain, including content, optics, batteries, chips, communications, and testing.
After more than a decade of waiting, the AR glasses industry has finally reached the doorstep of an IPO.
II. Can AI Save AR Glasses?
The biggest issue with AR glasses over the past decade has been that no one could clearly explain why consumers would wear them every day.
Over that decade, the industry's mainstream answer was optical display, using waveguide or Birdbath solutions to overlay digital information onto the real-world field of view, creating a "what you see is what you get" future computing experience.
But this path is inherently burdened by a heavy constraint: it’s difficult to simultaneously achieve a lightweight, slim design, long battery life, and high image quality while accommodating the light engine, display module, and optical components. As a result, AR glasses have never won over the general public—they’re novel enough to attract tech enthusiasts, but not practical enough to retain everyday users.
The industry has been waiting for a killer application until the emergence of large models, which have given AR glasses a soul. Rokid has also adjusted its product roadmap amid the AI boom.
In November 2024, Rokid officially launched its consumer-grade AI+AR glasses, the Rokid Glasses. Weighing only 49 grams, the device eliminates the bulky optical components of traditional AR equipment, retaining a minimalist display module designed to efficiently deliver lightweight, high-frequency information such as speech prompts, real-time translation subtitles, and navigation arrows. It pairs with a dedicated smart ring for intuitive page-turning control.
Unlike the industry's extreme all-screenless approach, this product strikes a balanced compromise with a lightweight body and targeted practical displays, perfectly harmonizing everyday wearability with functional utility.
What truly enabled the product to reach a broad audience was a public exposure event.
On February 18, 2025, at the High-Quality Economic Development Conference in Yuhang District, Hangzhou, Rokid founder Zhu Mingming delivered a speech entirely from memory while wearing Rokid Glasses. The full script was displayed in real time on the lens screen, and page turns were controlled effortlessly with a ring. No printed notes or低头 prompting were needed—this intuitive, practical application made the real-world value of AI glasses instantly understandable to the general public.
The video of Zhu Mingming delivering an impromptu speech while wearing Rokid Glasses briefly went viral, enabling Rokid to break out of its industry niche and achieve a critical leap in public awareness.
Traffic and popularity quickly translated into tangible market orders, with demand far exceeding the company’s internal projections. At the beginning of the year, Rokid’s internal forecast for annual shipments was only 100,000 to 150,000 units. However, official data shows that in 2025, Rokid Glasses sales surpassed 300,000 units—three times the original target.
At CES 2026, Rokid further evolved its product logic and launched the all-new screenless AI glasses, Rokid Style. Weighing just 38.5 grams, the product completely eliminates the display module, returning to a pure AI perception and voice interaction approach. Powered by a dual-chip architecture and a 12MP Sony sensor, it emphasizes all-day wearability and 4K recording, directly competing with Meta Ray-Ban in the overseas market.
Seen in the context of the industry, this is not just Rokid’s story.
According to Omdia data, global shipments of AI glasses increased by 322% year-over-year in 2025, reaching 8.7 million units. More notably, the user base has expanded to include a broader range of professions, such as business professionals, civil servants, lawyers, and teachers.
As a category of hardware products gradually reaches the mainstream, the true turning point for AI glasses is just beginning.
Three, the Battle of a Hundred Mirrors
At this point, the story should have ended on a satisfying note. After a decade-long freeze, general-purpose AI large models have filled the long-missing use cases for AR glasses; leading players like Rokid have embarked on their paths to capitalization, and the commercial narrative for the next generation of human-computer interaction interfaces is already taking shape.
But the true industry elimination contest has fully commenced precisely at this turning point, as the nationwide "Hundred Mirror War" officially kicks off, sweeping across both domestic and international markets.
Xiaomi has secured the top position in domestic AI glasses shipment volume for 2025 through its ecosystem strategy. According to the latest data from market research firm Counterpoint, Xiaomi ranked first in the Q1 2026 domestic smart glasses shipment chart with a 28% market share.
Huawei officially launched the HarmonyOS AI glasses in April this year, leveraging HarmonyOS's cross-device collaboration capabilities to capture the business and consumer markets; Alibaba introduced the Qwen AI glasses, deeply integrated with its life services ecosystem; and ByteDance has also planned to launch the Doubao AI glasses.
Xu Wu, partner at ZhenFund, believes that domestically, smart glasses have not yet moved beyond the “brand-driven” stage—primary users are fans of the brand itself, rather than purchasing due to essential AI-powered features, but instead out of trust in the brand or because they are already part of the brand’s ecosystem.
This is precisely the inherent weakness of startups like Rokid. Giants possess hundreds of millions of existing users, mature distribution channels, and strong supply chain bargaining power, enabling them to quickly replicate similar products and neutralize first-mover technological advantages. Meanwhile, Rokid has long focused on technology and product development, but its brand recognition among the general public and ecosystem barriers are still being built, causing its first-mover advantage to shrink.
Greater pressure in more competitive overseas markets.
According to Omdia’s 2025 Global AI Glasses Market Report, by shipment volume, Meta holds an 85.2% global market share with its Ray-Ban smart glasses, surpassing the combined sales volume of all Chinese-exported smart glasses brands.
At the same time, the onset of industry capitalization has exposed the real profitability challenges facing the AI glasses sector.
XREAL's prospectus shows that its revenue for 2023, 2024, and 2025 was RMB 390 million, RMB 394 million, and RMB 516 million, respectively, with gross margin increasing from 18.8% to 35.2%, nearing Apple's hardware business gross margin. However, XREAL still incurred a net loss of RMB 456 million in 2025, with cumulative losses exceeding RMB 2 billion over the three years.
This also means that, at this stage, even as product revenue and gross margin improve in tandem, the industry as a whole remains in an investment phase focused on gaining scale at the expense of profitability. Rokid has not disclosed any profit data to date, and whether it is profitable remains unresolved. We believe its ability to achieve profitability still depends on its capacity for scaling.
Zhu Mingming has set clear sales targets for Rokid: over 1 million units in 2026, 2 to 3 million units in 2027, and more than 10 million units in 2028.
In the consumer electronics industry, annual shipments of 2 million units are typically seen as the critical threshold for establishing a foothold in the mainstream consumer market. This means Rokid must transform its current short-term hype into sustained consumer demand over the next two years, relying on steady sales to amortize R&D and supply chain costs, and successfully transition from a niche product to a mainstream one.
Twelve years of patient waiting have given Rokid a ticket to the game, but also brought it into a far more competitive arena. Moving forward, Rokid must prove that a startup can become a truly cyclical-resistant company amid giants.