Alibaba's Ambition That Cannot Be Contained

Author | Chai Xuchen

Editor | Huang Yu

"In the future, every pair of glasses will be AI glasses. Choosing Quark AI glasses is equivalent to choosing Alibaba's Qianwen large model and the entire Alibaba ecosystem."

On November 27, Alibaba Group Vice President Wu Jia endorsed Quark AI glasses, revealing Alibaba's ambitions in the era of artificial intelligence. On this day, Quark unveiled its products, launching six items across two series, S1 and G1, all equipped with Alibaba's latest Qianwen AI assistant.

Both series feature numerous functions such as navigation, payment, and AI Q&A, aiming to become a true portable super assistant. At the same time, Quark AI glasses have integrated Alibaba's internal ecosystem, covering vertical usage scenarios such as search, navigation, payment, and business travel, breaking through the ecological dilemma currently faced by AI glasses.

Wall Street Insight experienced firsthand that the Quark S1 glasses can smoothly respond to voice assistants in noisy environments; quickly compare prices by glancing at products; and confirm payment by scanning codes with just a gaze, all without touching a phone.

As Wu Jia mentioned, one of the "unique skills" of Quark AI glasses is the great integration of the Alibaba ecosystem, supported by a range of Alibaba applications such as Amap, Alipay, and Taobao.

Quark AI glasses seem to be arriving at the right time, as the industry is entering a "battle of hundreds of glasses." According to data from the National Bureau of Statistics, from January to October this year, the national online retail sales increased by 9.6% year-on-year, with online retail sales of smart wearable products like AI glasses and smartwatches growing by 23.1%. During "Double 11," the transaction volume of AI glasses on Tmall surged by 2500%, while the same category on JD.com saw a transaction growth rate of 346%, topping the 3C digital list.

However, industry insiders remain curious as to why Alibaba, already a leader in e-commerce, AI, and cloud services, is venturing into smart hardware.

"Because Alibaba has always faced a dilemma: where does the traffic come from?" an executive from the AI glasses sector told Wall Street Insight. "The reason Alibaba has always wanted to engage in social media is that the most active user behavior occurs in social settings, while Taobao, Amap, Alipay, and others are tools that are used and discarded immediately. Therefore, Alibaba is eager to control traffic itself."

Where is the primary entry point for user traffic? Years ago, during the mobile phone era, Alibaba attempted to create a cloud OS and cloud phone, but these efforts were unsuccessful. They even acquired a mobile phone manufacturer and later developed Tmall Genie, all of which were explorations of Alibaba's primary entry point for users.

According to the aforementioned executive, mobile phones and smart speakers were once entry points, but now it is smart glasses, backed by AI OS. "Alibaba's pursuit of this indicates that everyone recognizes smart glasses as the next interactive computing terminal and new entry point, and that this entry point combined with AI brings even greater disruption."

In a post-event discussion, Song Gang, head of the smart terminal business of Alibaba's Intelligent Information Business Group, pointed out to Wall Street Insight, "We believe that AI glasses have the potential to become the 'sensory hub' of the next generation of human-computer interaction, and are the devices most likely to challenge mobile phones in the future." Moreover, in today's AI era, glasses have more advantages than smartphones. Wu Jia believes that AI glasses are the intelligent devices that truly open the revolution of human-computer interaction in the AI era, and their importance is no less than that of smartphones. For example, as a first-person perspective head device, AI glasses can capture over 80% of human sensory input, something smartphones cannot achieve, allowing Qianwen to fully connect with the physical world.

In Song Gang's view, "Glasses do not display all the applications like today's smartphones; users pay, navigate, and awaken 'Qianwen', with all third-party services hidden behind the intelligent assistant." This is precisely what Quark's head Wu Jia refers to as the meaning of "All-in-One" and "Doraemon's pocket."

This "pocket" is Quark glasses; the treasures in the "pocket" are Qianwen and all the encapsulated Alibaba ToC capabilities.

Song Gang said, "As we transition from the past OS ecosystem to AIOS, and the current apps are being transformed into agents. In the past, you might have to download an app yourself, open it to search and use it; this process can now be completed with a single sentence on the glasses. This is more conducive to the scenarios for glasses, which is the main direction of technological development."

Industry insiders have bluntly stated that in the future AI era, you will find that there are no apps anymore; the current AI assistants or voice interaction agents will serve as a unified entry point to help you execute tasks. The future entry points will be more centralized and unified, "this is quite frightening; the future AI operating system may turn large models into operating systems."

In the past year or two, all of Alibaba's changes seem to be paving the way for this grand future. Imagine if Alibaba had not returned to the big group strategy from "one split into six," how could it mobilize Alibaba Cloud, Amap, Shansong, and Quark to support Qianwen's new battlefield? Without the ALL IN AI grand strategy, Alibaba might still be just an e-commerce company.

At this moment, Qianwen serves as the front-end interface, thoroughly re-packaging and unifying Alibaba's past ToC capabilities scattered across various business lines, making a full sprint towards the AGI era.

The current capital market is very realistic; what they care about is whether the powerful AI parameters and capabilities can ultimately be monetized.

In this regard, Alibaba is ambitious. In the third fiscal quarter of this year, Alibaba Cloud's revenue surged by 34%, and AI products have seen triple-digit growth for nine consecutive quarters. After showcasing its B-end muscle, the C-end will be the next big show. If this logic can be validated, it will prove that Alibaba's AI strategy also has the ability to continuously generate revenue on the C-end, and Qianwen will bring greater imaginative space to Alibaba.

Below is the transcript of the conversation with Jin Xian, head of Alibaba's smart hardware, and Song Gang, head of the intelligent terminal business of Alibaba's Intelligent Information Group:

Q: In what directions will Quark AI glasses expand their application scenarios?

Song Gang: Today, we see that the Alibaba ecosystem has already been integrated and deeply customized, supporting navigation, payment, Taobao Instant Purchase, and Fliggy. We will also continue to expand into sports, health, and learning scenarios.

Q: How much resources and patience is the entire Alibaba system currently investing in this AI glasses? Song Gang: It's definitely not a small investment; Alibaba understands this very clearly. Wu Jia has also articulated the strategic positioning well. We believe it is the next generation of personal mobile entry and also the entry point for AI, the center of human-computer interaction revolution, and the device that has the greatest potential to challenge smartphones in the future. In Alibaba's strategy, AI glasses play a very important role.

Today, when we create hardware, the investment is actually comprehensive. Alibaba has full-stack technical capabilities, and we have designed the hardware in all aspects, including specifications, comfort, and aesthetics. Coupled with the support of the Qianwen large model and Alibaba's own ecosystem, this is not something that can be achieved by simply making a piece of hardware. There are very few companies that can bring all of this together.

Question: How will AI hardware evolve in the future? Major companies claim to have ecosystem advantages; what can they actually achieve?

Song Gang: Glasses will definitely be the main character. As we transition from the OS ecosystem to AI OS, and as current apps are transformed into agents, the past process of downloading apps, opening them, searching, and using them can now be completed with a single sentence on the glasses. This is more conducive to the scenarios for glasses, which is the main direction of technological development.

AI OS will lead to significant changes in the hardware industry in the future. Today, we still need to wake up Qianwen; tomorrow, we won't need to wake it up, and combined with eye movement, such control will also be relatively smooth.

The development trajectory is still quite clear. Additionally, AI's ability to process massive amounts of data will be something that past devices did not possess. Glasses may be equipped with more sensors, leading to a deeper understanding of people and the environment. Based on the processing of massive data, they will be able to serve people more proactively.

Question: How is the computing power of Quark glasses distributed and coordinated between the edge and the cloud?

Jin Xian: Glasses must be lightweight; they won't replace smartphones all at once but will gradually penetrate smartphone scenarios. If smartphones used to be used for 6 hours a day and now only for 5 hours, that extra hour has gone to your glasses. So today, the computing power of glasses still relies on smartphones, even depending on the cloud on smartphones, but in the future, it may bypass that.

We are still more reliant on the cloud in the entire AI ecosystem. I have tried many AI glasses, and it takes about 7 seconds to answer a question. For us, we can optimize that to within 3 seconds. This difference is because we can collaborate from the edge to the cloud and to the mobile end.

Question: What is the content and ecological thinking behind Quark's AI glasses?

Song Gang: Besides the Alibaba ecosystem, we have also integrated external resources such as NetEase Cloud Music, QQ Music, and Flight Manager. In the future, there will be a principle that we will prioritize integrating resources that are more suitable for the glasses scenario. Looking ahead, we are also building an ecosystem for developers.

Question: Why did you choose the combination of Qualcomm and Hengxuan for computing power?

Jin Xian: We adopted a dual-chip, dual-system approach. AR1 runs Android, while Hengxuan runs the RTOS real-time system. These two chips are specifically designed for glasses. Our architecture can lead for quite a while because both sides are working hard to drive iteration. In the past, smartphones also had dual chips, and at that time, they were the best smartphones available Question: Why not directly use full-color screens in this generation?

Song Gang: Before the emergence of AI glasses, AR had already developed for more than a decade. Why has there not been a significant leap in quantity? To some extent, it can only be worn for 20 or 30 minutes.

The definition of AR itself should not be immersive; it should be a combination of virtual and real that cannot detach from reality. Therefore, it's not that we can't do it with the current technology, but can we do it better?

There is no problem with this direction, but there are better alternatives for viewing, which tend to be more immersive. If we take this path towards entertainment, we will lose transparency to reality. We believe that this wave driven by AI should first develop towards tool-oriented, helping, and assisting people.

We believe that AI glasses should be worn all day. The current color light machines have not yet reached a state where they can be worn out at any time. We are already researching better color light machines, which will be the industry's first to launch smaller, brighter, and better-experienced color light machine products, but it will take time.

In the future, whether seamless switching between immersive and non-immersive experiences is possible depends on the technology of the light machine itself. Even the light machines currently under research cannot provide a very good viewing experience; they are still information screens that can better combine virtual and real elements.

Question: Other AI glasses have also integrated the Tongyi large model. What are Quark's advantages?

Jin Xian: We are a company that originally focuses on AI business. From Quark to Qianwen model, we have accumulated many years of experience in reverse-defining models, and many of Quark's tool usage scenarios are highly compatible with the user scenarios of glasses today.

Today, glasses primarily solve the usage problems in tool scenarios. Our past business accumulation, model definition capabilities, and the design of the entire engineering chain actually point to the fact that the glasses and models we are working on today are absolutely highly matched.

Question: What bottlenecks need to be overcome for AI glasses to transition from a gimmick to practical use in the future?

Song Gang: AI will go through three stages: learning from humans, assisting humans, and surpassing humans. The products released today have already reached the stage of assisting humans; they are definitely not toys, at least they are tools.

To reach a level of usage that can replace smartphones, we indeed face a revolution in materials, such as breakthroughs in higher transmittance glass, silicon carbide materials, and high-density batteries. The trend of building an industrial chain for glasses also relies on miniaturization. The excess space of this product will be given to its appearance; the current industry bottleneck is still more in materials and chips.

Additionally, there is the breakthrough of ecology. Whether it can develop vigorously like the mobile phone ecosystem also heavily relies on players like Alibaba to open their ecosystems and connect data.

Question: What future functions can be co-created with the supply chain?

Jin Xian: Internally, we place great importance on competitiveness, which we call architecture. One is the overall machine architecture, and the other is the software and hardware system architecture. Today's overall machine architecture can support what I just mentioned at the press conference, called sensory priority and front-and-back collaboration. In fact, this is something that no one in the industry has thought about in terms of how to build its architecture To achieve this architecture, it is necessary to customize with supply chain partners such as optical machines, waveguide plates, batteries, and FPCs. Often, the difficulty of customization exceeds the normal shipping capabilities of the original supply chain, which can achieve good yield rates. Therefore, we have done a lot of innovative work with the supply chain to bring many yield rates to a state where they can be shipped in large quantities today.

The second aspect is the hardware and software system. We have integrated Qualcomm and domestic chips to work together, along with the design of battery swapping. Thus, turning a DEMO into something that can be used in daily life involves a huge amount of work. The second point is perception; it must assist people and have stronger perceptual capabilities, which is also a direction we need to work hard to overcome.

Q: What conditions should future AI glasses in China add to achieve a leapfrog advantage?

Song Gang: First, look overseas. META indeed relies on traditional glasses channels for its design, especially its classic models, which can sell in large quantities. By adding a few dozen dollars, it becomes smart glasses. Even if it is not that smart, it is still a branded pair of glasses. Its business model relies more on traditional eyewear brands and channels to complete the commercial loop, and the AI capability is relatively weak, especially in the domestic market where it is not very usable.

I believe that in the future, our model, driven by AI and combined with traditional eyewear, will be more AI-driven. Our design will also lean towards enabling long-term use, aligning with the trend of APP agent transformation.

The overseas company also wants to find one or two more brands. With such large single model volumes, it is also difficult for them to move up and will return to this path. Domestically, we are already achieving a leapfrog advantage; although the volume is not that large yet, our designs have already surpassed everything comprehensively.

Q: After the battle of the hundred glasses, which types of manufacturers are more likely to survive?

Song Gang: The current industry is somewhat like the era of feature phones, with a hundred flowers blooming and each having its own definition. Today, looking at this market, the demand still far exceeds what has been met, and there is still a lot of growth potential. With the progress of AI, the development of hardware technology, the maturity of the supply chain, the continuous decline in prices, and the constant improvement of functions, the imaginative space is actually very large.

Each manufacturer needs to find its own positioning and establish a closed loop where its business logic can hold. All of them can survive because the market is large enough. This year, the global mobile phone market is about 1.24 billion units, but the global eyewear market exceeds 1.5 billion units, and the replacement cycle for glasses is shorter because it is coupled with fashion factors, leading to a faster iteration cycle