（By Observer Net, Lv Dong, Editor/Zhang Guangkai）
This year, the most popular product in the consumer electronics industry is AI glasses. First, at the beginning of the year, Zhu Mingming from Rokid said "the speech is in my glasses," which went viral. Then, Xiaomi and Alibaba's subsequent entry added more heat to the popularity of AI glasses.
But in the midst of the industry's noise, AR lens company XREAL has not released its own AI glasses for a long time. According to data released by IDC, XREAL has been the global sales champion in the AR field for three consecutive years. In the first quarter of 2025, XREAL ranked second in the global AR/VR industry, just behind Meta, surpassing internet giant ByteDance.
Regarding topics such as AI, chips, and operating systems that have attracted external attention, Observer Net recently had an exclusive interview with Wu Kejian, co-founder and head of algorithms and AI at XREAL, at the World Artificial Intelligence Conference. He admitted that the current AI glasses on the market are still relatively simple in function, mainly used for photography or as Bluetooth earphones, which is more like "multimedia glasses" rather than "AI glasses."
"These products lack core advantages. The supply chain has become highly mature, and one can easily make an AI glasses by integrating the supply chain of smartwatches or smartphones. The easy entry of big companies proves that no deep accumulation is needed to launch a product. However, this model is not the core value of XREAL. It is far from our expected fully intelligent form, and we will not take this path," Wu Kejian said.
XREAL was founded in China in 2017 by Xu Chi and Wu Kejian, among others. They all returned from the United States to start their businesses.
As of now, XREAL has nearly 500 employees and has raised over $300 million in funding. Investors include Alibaba, Kuaishou, NIO Capital, Yunfeng Fund, Sequoia, and Hillhouse. Unlike some AR startups, XREAL not only designs glasses but also focuses on manufacturing. They have established two optical factories in Wuxi and Kunshan, with 65% of components customized and produced in China.
"Over the past decade, smartphone manufacturing has allowed China to accumulate significant industrial chain advantages. If XREAL were established in the U.S., the manufacturing cost would be several times higher than it is now," Wu Kejian candidly told Observer Net. "Relying on China's hardware manufacturing advantages, from the brand and terminal perspectives, we can completely build a world-leading enterprise."
Whether it's AR glasses or AI glasses, the industry's turning point is clearly not yet here. Wu Kejian admitted that hardware still needs to break through bottlenecks such as optical display and battery life, and it will take five years or more to reach the ideal form. But he firmly believes that the size of glasses is completely comparable to that of phones. "People will not completely do away with phones when they wear glasses; it will definitely be a coexistence concept."
Recently, XREAL became the first and only Chinese company to collaborate with Google in the global AR field. Project Aura will be the world's first flagship smart glasses powered by XREAL's self-developed X1S chip, specifically designed for the Android XR platform.
"There is competition and cooperation between China and the U.S. in AI development. We will see that technological innovation and product iteration are progressing in tandem. The collaboration between XREAL and Google is a good example. Google has software system advantages and ecosystem influence, while XREAL has years of hardware experience in smart glasses. Both sides are very willing and encourage such strong partnerships," Wu Kejian believes. "With the combination of China's industrial chain and global innovation, Chinese enterprises have the potential to define the next generation of terminals."
Wu Kejian, co-founder and head of algorithms and AI at XREAL
Here is the full transcript of the conversation:
Observer Net: This year, smart glasses have become a hot sector. Recently, Google released the Android XR system, which could solve the problem of missing application ecology in previous AR devices. Apple's recent release of Vision Pro also sparked the concept of spatial computing. How are Chinese companies doing in this sector? Today, we have invited Wu Kejian, co-founder and head of algorithms and AI at XREAL, to discuss the present and future of smart glasses.
Wu Kejian: Okay. I am Wu Kejian, co-founder of XREAL, responsible for algorithms, AI, and projects related to collaborations with companies like Google.
Observer Net: We have seen that XREAL recently launched two new products that have drawn a lot of attention: XREAL One Pro and Project Aura. Can you introduce these two products and their innovations? What are the differences from previous models?
Wu Kejian: Okay. Since its establishment, XREAL has focused on the fields of AR and AI glasses. This year, these two products represent our two main product lines: spatial display and spatial computing.
At the World Artificial Intelligence Conference, we mainly introduced the XREAL One Pro series, which is currently our flagship product in the consumer-oriented spatial display product line. Its function is as an AR glasses, allowing users to connect various devices such as mobile phones, computers, and game consoles. Once connected, users will have a private home cinema, where content from these devices is directly displayed in front of them, creating a similar experience to a home theater, and enjoying it anytime, anywhere.
This year, we made two core functional upgrades on the XREAL One and One Pro series: optics and chips. In terms of optics, the XREAL One Pro is equipped with our latest optical system. The advantage of this system is that it makes the glasses thinner while expanding the field of view, reaching an industry-leading level of 57 degrees. This means that when users watch the screen, the virtual screen in front of them is larger, as if they are in a large home theater or cinema.
While making progress in optics, we also upgraded the chips. All XREAL One series products are equipped with our self-developed X1 chip. This chip is a space co-processing chip specifically designed for AR glasses, primarily solving various connection compatibility issues and reducing the delay of spatial display effects to an industry-leading level of 3 milliseconds. Therefore, when people see the XREAL One series glasses, including comparisons with our previous products and other similar AR glasses, they will find obvious differences.
You also mentioned the collaboration with Google on Project Aura. In May, at Google's annual largest developer conference in the Bay Area—Google I/O—Google announced the collaboration project with XREAL, known as Project Aura. This is a new collaboration project of Google's Android XR, a set of operating systems for XR devices under the Android framework. Android XR is specifically tailored for XR devices, covering VR, AR, and AI glasses. Google hopes to create a universal operating system for XR devices, similar to how Android systems were popularized in the smartphone era. As Google's "Alpha Partner," XREAL has launched corresponding AR glasses, focusing on spatial computing capabilities. This device was officially announced at the I/O conference and is expected to be available early next year.
Observer Net: These two products seem to reflect XREAL's focus on R&D. In terms of hardware, you seem to focus more on refining the hardware experience; in software ecosystems, you tend to collaborate with giants like Google. Could you elaborate further on your collaboration with Google and future plans?
Wu Kejian: The XR industry has been developing for many years, but the entire industry is still in a highly fragmented state. Each company launches its own hardware products and installs its own system or software, which brings some challenges to the industry's development. It's normal for this phenomenon to appear in the early stages of the industry, but from the perspective of content development, developers find it difficult to develop an application that can work on multiple platforms or hardware devices. For example, an application developed for Meta Quest can only run on the Meta Quest, and an application developed for Apple Vision Pro can only run on the Vision Pro. Other manufacturers are also the same. Now, the hardware in the industry is developing well, whether it's heavy display forms, light displays, or even no display AI glasses, the hardware is rapidly improving. At this point, the industry needs a certain degree of unification in ecological or system platform capabilities to address the fragmentation issue.
Google is trying to solve this problem. Google has been working on developing the Android XR operating system for many years. The system aims to be backward compatible with various XR and AI devices and provide a unified open platform for XR and AI developers, allowing developers to develop applications that can be used on various devices, thus improving or solving the fragmentation issue to some extent. In short, Google hopes to create an Android system for the new era of XR. As a startup company, XREAL has launched multiple hardware products in the past few years.
We also tried to build our own ecosystem, launching SDKs and operating developer communities. But honestly speaking, the appeal of a startup company is relatively limited. Globally, developers tend to rely on large platforms to carry their content. Therefore, when we saw Google starting to build the Android XR ecosystem, we contacted Google early on, deciding to combine XREAL's advantages in AR glasses hardware with Google's strong influence in the Android ecosystem, forming a strong partnership to jointly launch a globally sold AR glasses product. Thus, developers only need to develop a set of applications for the Android system, and in the future, they can use them on XREAL and other hardware platforms that Google expands. This is the goal of Project Aura.
Observer Net: AR glasses have been developing for many years, always facing the problem of lacking native content. After the launch of Android XR, will it be more inclined to develop native content?
Wu Kejian: There are mainly three types of content in Google's Android XR devices. The first type is native 2D smartphone and tablet applications, which will be directly migrated to Android XR devices and maintain compatibility. Although these applications have a similar visual experience to traditional 2D devices, the screen size will be larger. The second type is applications developed by Google itself, such as Google Maps, Google Search, YouTube videos, and Google Photos. These applications will be 3D and XR processed to demonstrate the 3D content effects that can be achieved in the Android XR operating system. The third type is that Android XR will support and be compatible with the mainstream XR open platforms known so far. Developers can directly develop applications based on Android XR or seamlessly transfer content based on the OpenXR standard interface to Android XR. Google has invested a lot in content ecology, trying to be compatible with existing 2D and 3D content as well as native content.
Observer Net: Will the ecosystem of smart glasses become as rich as the app ecosystem of smartphones in the future?
Wu Kejian: Indeed, it will. Currently, regardless of AR or AI devices, the entire industry is still in the early stage, so it is relatively fragmented, and there are fewer targeted applications. However, it is certain that the potential of this device form is huge, and all tech giants are actively investing in this field. With the increase in hardware numbers and the unification of system platforms, a good content ecosystem foundation will be formed, thus nurturing dedicated applications for AR and AI glasses. At that time, these applications will achieve a qualitative leap in interactivity, perception capability, and AI functions, surpassing familiar products like phones and computers.
Observer Net: In terms of software ecology, you have made a firm cooperation with Google. Then, in terms of chips, XREAL has recently iterated its self-developed chips. What is the consideration behind this? In the AR sector, there are few companies that develop their own chips.
Wu Kejian: In fact, we may be the only company among startups that is developing chips. Because we have made several generations of glasses, we found that there are some inherent problems with current AR glasses and AI glasses, which cannot be solved by the existing smartphone supply chain. For example, current smartphone chips are designed for smartphones. If their computing power is high enough, the power consumption may not be suitable for the development of AR and AI glasses, leading to poor battery performance. On the other hand, AR glasses have very high requirements for display, needing to support 3D spatial display. However, smartphone chips cannot support this native 3D display because smartphones are 2D screens, and their display requirements are completely different. From these perspectives, we found that there is no chip on the market that can fully meet the needs of AR glasses. Although Qualcomm is a relatively active player in this field, its chips are also designed for heavier computing platforms and are not suitable for direct application in lightweight AR glasses. Given this, although we are a terminal manufacturer, mainly responsible for hardware products, in the early stages of the industry, when the supply chain has not yet fully matured, we believe it is necessary to independently develop a chip to improve the overall performance of the product. This self-developed chip is not competing with Qualcomm and other existing chips, but has a different positioning.
In fact, this chip is an AR co-processor, not a main computing chip. It can be compared to Vision Pro, which has two chips: one is the M-series main application chip, similar to a laptop chip; the other is the R1 chip, specifically designed for XR devices. Our X1 chip follows a similar approach, a co-processor specifically designed for AR glasses, responsible for display, simple interaction, and spatial perception functions. It works in conjunction with main chips like Qualcomm, and the application system still runs on the main chip. Our self-developed chip mainly solves real-time processing issues related to AI glasses end, and has extremely high requirements for power consumption control, which is the positioning of this chip.
Observer Net: Your chip is mounted on the glasses, does this mean that some data no longer needs to be transmitted to other devices for processing?
Wu Kejian: This is a key issue. With this co-processor chip, the glasses can handle tasks that require low latency but have a light processing load. This reduces the need to transmit raw sensor data or images to a phone or computer, significantly reducing system power consumption and improving processing effectiveness because local processing reduces latency. From a technical perspective, this is very reasonable. Of course, developing this chip itself is also a challenge.
Observer Net: Apple's Vision Pro has a latency of 11 milliseconds, while yours is 3 milliseconds. What does this mean?
Wu Kejian: First, there is a clear difference between these two products. Apple's Vision Pro is a VST (Video See-Through) device, which uses high-definition cameras to capture the outside world and display it on the screen. Its latency is 11 milliseconds, which is already quite excellent in the industry. Our AR glasses use OST (Optical See-Through) technology, allowing users to see the outside world through prisms without additional capturing and displaying steps. Therefore, compared to Vision Pro, our product has relatively lower computational and chip performance requirements, achieving a latency of only 3 milliseconds.
Observer Net: Does the reduction in latency bring improvements to actual user experience?
Wu Kejian: In the industry, 10 milliseconds is usually considered the standard. When the latency drops below 10 milliseconds, even to 5 milliseconds, users almost cannot perceive the image delay. This means that when users use AR glasses, the experience will be almost indistinguishable from observing the real physical world. The stability of moving images is significantly improved, and users will not feel image jitter or dizziness, resulting in a more comfortable user experience.
Observer Net: At this year's artificial intelligence conference, the concept of AI glasses remained hot. XREAL has led the global sales for several years, but in terms of AI, it seems not to be aggressive. Is it because internal progress has not been disclosed, or is there another consideration?
Wu Kejian: Since the rise of AI glasses, we have always closely watched the industry dynamics. However, to date, we have not launched a product similar to Ray-Ban Meta. Many domestic manufacturers have already launched AI glasses, and even the term "hundred glasses battle" has emerged. We did not rush to follow suit, and we will not launch such a product this year. The reasons are as follows: First, the AI attributes of current AI glasses are still weak. Although many products are connected to large models, the proportion of actual AI functionality used is extremely low, and the experience is poor. Second, the functions of most glasses on the market are still relatively simple, mainly used for photography or as Bluetooth earphones. I prefer to call them "multimedia glasses" rather than "AI glasses." These products are not what XREAL pursues, nor are they our strength. What we truly want to pursue is the ultimate form of smart glasses: in addition to AI, cameras, and voice, they must also have three core AR capabilities: display, interaction, and spatial perception. Without these enhanced reality features, glasses can only complete basic tasks such as taking photos, scanning, listening to music, or making calls, far from the comprehensive intelligent form we expect. Therefore, we are not rushing to launch such products. For many companies, especially startups, these products lack core advantages: the supply chain has become highly mature, and one can easily make an AI glasses by integrating the supply chain of smartwatches or smartphones. The entry of companies like Xiaomi proves that no deep accumulation is needed to launch a product. This model is not the core value of XREAL, and we will not take this path.
Observer Net: How will you differentiate yourself in the future?
Wu Kejian: We have a clear plan for the evolution path. Whether called AR glasses or AI glasses, the final form must have a unified set of functions. In addition to AI, photography, and music, they must integrate display, interaction, and spatial perception. When worn, the glasses can perceive the environment, provide visual feedback information, and support multimodal interaction, which is a complete consumer-end smart device. Therefore, we do not make current simple AI glasses but directly invest in the next stage. At that time, people will see that it includes all the capabilities of the current simple form of glasses, and achieves a significant enhancement on top of that. Essentially, it is not two paths, but the same direction of evolution; however, we are not interested in the current simple stage and choose to move directly into the next stage.
Observer Net: In the industry, some companies have divided the smart glasses route into two paths: AR glasses and more lightweight AI glasses. Will you do the same in the future?
Wu Kejian: In the next few years, the market will indeed see various forms: one that focuses on all-day wear with relatively weak functions; the other that focuses on immersive viewing, with high picture quality requirements, but cannot be worn all day. Smart glasses have a wide range of applications, meeting daily needs, immersive scenes, or professional and office-specific tasks. In the next few years, the market will see various forms of smart glasses: some manufacturers focus on immersion and display effects, some focus on all-day wear, and some focus on professional scenarios. Due to the broadness of the sector, each company's niche positioning is already very clear, so you will see many different hardware forms. However, in the long run, these forms will eventually converge, achieving a balance between wearing comfort and functional completeness.
Observer Net: Currently, companies like Xiaomi, Alibaba, and Baidu have entered the glasses market. They have greater financial, technological, and resource advantages. As a co-founder of XREAL, how do you view their entry? Will it bring challenges?
Wu Kejian: The products launched by big companies are all simple form AI glasses. The reason is as I mentioned earlier: the supply chain has become highly mature. If a company were to start today, it could quickly launch hardware without deep research and development, then cooperate with domestic AI large model companies. Therefore, the entry of big companies indicates that the supply chain for simple form AI glasses is ready. Second, the reason big companies have entered is that they recognize this industry and see its enormous development potential, otherwise they would not have invested. These two points confirm the current situation and prospects of the industry. Of course, looking at the current situation, everyone is still doing simple form AI glasses. To move to the next stage and achieve more complete smart glasses, core aspects such as chip technology, optical display technology, battery technology, and the overall software experience of AI glasses need to be broken through, which still requires time, and I hope everyone will remain patient. From past experience, such cutting-edge breakthroughs are often led by vibrant startups, and XREAL is one of them.
Observer Net: Startups' advantage lies in focusing, while big companies have too many areas to concentrate on?
Wu Kejian: I think there are two main points. First, firm belief. Big companies can do this today and that tomorrow because they already have very profitable main businesses. XREAL has been focusing on one thing for eight years since its establishment—perfecting AR glasses and the underlying optical, chip, and display technologies. Second, our organization, process, and way of doing things are centered around AR-native. For example, this is completely different from a company with an internet APP-centric gene. The same task will be handled differently in two companies.
Observer Net: What areas are you currently focusing on?
Wu Kejian: Mainly two areas. One is to continue advancing the iteration and productization of XREAL's algorithms and AI, with each new product containing the entire algorithm and hardware-software collaborative achievements. The second is serving as the project leader for the Google Project Aura collaboration, dedicating a considerable amount of energy to this project.
Observer Net: The development trend of AI capabilities in phones is end-cloud collaboration. Will smart glasses follow the same trend in the future?
Wu Kejian: Smart glasses require higher AI functions, needing large models, multimodal capabilities, and strong memory abilities, but they must also be thin, low-power, and long-lasting. Therefore, I believe end-cloud collaboration is an inevitable trend.
Observer Net: Has XREAL already accumulated some capabilities in end-side AI?
Wu Kejian: We have long focused on end-side AI, with all algorithms and AI work centered around the glasses' end-side capabilities. Our self-developed X1 chip has already run several specialized small models, specifically for image processing, interaction recognition, and other specific tasks. Our algorithm and AI team continues to work hard, and subsequent products and features will continue to expand.
Observer Net: Market research institutions show that XREAL has consistently ranked first in global sales for many years. Facing overseas giants like Meta, what do you think are the advantages of Chinese companies?
Wu Kejian: The advantages are mainly in several aspects.
The first is the industrial chain advantage. Myself and Xu Chi (co-founder and CEO of XREAL) both returned to China from the United States to start our businesses. First, because we are Chinese, but more importantly, because making hardware must rely on a very good hardware supply chain and manufacturing industrial chain. China has a leading position in the world in this regard. Over the past decade, smartphone manufacturing has allowed us to accumulate significant industrial chain advantages, so we found that making hardware in China is much cheaper and more efficient than abroad. If we were to establish XREAL, an AR glasses company, in the United States, the manufacturing cost would be several times higher than it is now.
The second is the regional ecological advantage. Focusing on Shanghai and the Yangtze River Delta region, the AR industrial chain is complete, and we can find high-quality enterprises in the upstream and downstream key links. Recently, we have moved our headquarters to Pudong, and we appreciate the support from the Pudong government. In the future, we will rely on the industrial advantages of Shanghai and the Yangtze River Delta region to promote the company's further development.
The third is the market advantage. In addition to the industrial chain, the Chinese market is also crucial. XREAL is rooted in China, and its products are sold globally. The Chinese market has always played an important role. Over the past decade, the popularity of smartphones and the Internet has cultivated a vast consumer electronics potential, and Chinese consumers are highly interested in and accepting of new smart devices. The activity of AI glasses and AR glasses in China is no less than that in the U.S. and other overseas markets. Therefore, China has always been a very important market for us, and in the current international competitive landscape, the Chinese market remains a very important market for us.
Observer Net: The smartphone era was led by the iPhone. Now, China has a mature industrial chain and a vast consumer market, and technological innovation is constantly breaking through. Do you think Chinese entrepreneurs can define the next generation of terminals?
Wu Kejian: I have said many times that in the PC and smartphone eras, China missed the first-mover advantage, and Western companies defined these hardware and operating systems, resulting in many bottlenecks and making us feel uncomfortable. Therefore, we should catch up or even lead in the next generation of smart hardware. I believe the next most promising one is smart glasses. Currently, in the smart glasses category, China is in a relatively advanced position. Relying on China's hardware manufacturing advantages, from the perspective of brands and terminals, we can completely build world-leading enterprises.
Observer Net: China's industrial chain has unparalleled advantages. Can we combine China's advantages with global innovation resonance?
Wu Kejian: That's very correct. For example, Google is an American tech giant, and we are a Chinese startup. This joint effort also reflects that the AR and AI industries are still in the early stages of development and are not yet fully mature. At this stage, technological innovation is borderless and not restricted by geopolitical factors. We have found that technological exchanges between China and the U.S. are very frequent. China and the U.S. have competition in AI development, but there is also cooperation. However, we have found that the entire technological innovation and product iteration are progressing in parallel. The collaboration between XREAL and Google is a good example. Google has software system advantages and ecological influence, while XREAL has years of hardware experience in smart glasses. Both sides are very welcoming and encouraging of such strong partnerships. I think China and the U.S. have a relationship of both competition and cooperation in these new categories.
Observer Net: When will the turning point for smart glasses arrive? How far is it from the state that satisfies you?
Wu Kejian: Many people say that this year is the year of AI glasses, and I think in a way, it's correct. This year, the AI glasses industry has been very exaggerated, possibly having 100 brands of AI glasses. People have seen all kinds of products, and many people have started to get involved and actually buy and wear such products. This phenomenon has never occurred before, which is a turning point. But in the opposite sense, this is not the product we are satisfied with, at least not the product that XREAL is satisfied with. For most consumers, after experiencing the product, they will also feel that it's a bit lacking. So I think the real turning point has not come yet. Now, AI glasses are indeed very hot, and many people are willing to get involved and try, but I think the distance to the form we want is still a long way.
Observer Net: What challenges still need to be addressed?
Wu Kejian: Let's not talk about software. Hardware still has many challenges, including optical display, chip technology, and battery life technology, etc. We ourselves are in the industrial chain and technology end, and we still need many frontier breakthroughs. This is not something that can be solved in one or two years, so everyone needs some patience, and there will be some good phased products. I encourage everyone to try and experience it, but you also need some patience. The distance to the state we ultimately want may still require five years or more.
Observer Net: But you are confident that glasses will become a best-selling product like phones?
Wu Kejian: Very sure. We believe that the scale of glasses is completely comparable to that of phones. Because glasses are definitely a personal computing-level consumer electronics. So from this angle, you can imagine that in the future, when you wear glasses, you may still need a phone, not completely getting rid of it. It is definitely a coexistence concept. When there are glasses, in many cases, you can just use the glasses, and you don't necessarily need a phone, just like the relationship between phones and computers today. So glasses will definitely become an important part of people's lives, and the time spent on them may not necessarily be less than that of phones.
Observer Net: But just like the long development process of smartphones, the experience of smart glasses also needs to be gradually improved.
Wu Kejian: Now, hardware still needs many technological breakthroughs, and software systems have just started to have some system-level or platform-level ones. Everything is just starting to develop.
Observer Net: Alright, thank you, Dr. Wu Kejian. The conversation today is now concluded.
Original article: https://www.toutiao.com/article/7536103888626139683/
Statement: This article represents the views of the author and is welcome to express your attitude via the 【top/down】 buttons below.