Interview with Li Hongwei of Thunderbird Innovation: AI glasses without a display are just transitional products.

"Don't put all your eggs in one basket." This age - old wisdom in life can also summarize the strategic thinking of Thunderbird Innovation.

The biggest problem that AI/AR glasses face today is no longer the bottlenecks in technology and the supply chain. Instead, technologies such as optical engines, screens, and large - scale models are advancing too rapidly, and industry practitioners are easily dazzled by the plethora of options.

(For a case where excessive technology stacking led to product failure, refer to the previous article in Intelligent Emergence: An AR company invested by Alibaba collapsed at the peak of the AI glasses trend)

For Li Hongwei, the founder and CEO of Thunderbird Innovation, how to translate technological reserves into the optimal product solution is an extremely crucial proposition.

Thunderbird Innovation has a wider and deeper range of available technology stacks. Founded in 2021 and incubated by TCL Electronics, Thunderbird Innovation's privileged background means that it has an edge over startups established during the same period in terms of technologies such as optics and display.

However, with abundant technological reserves, it is even more necessary to learn restraint.

"We won't pile all the most advanced technologies onto the product just to show off our skills," Li Hongwei told Intelligent Emergence.

In product definition, Thunderbird Innovation applies the philosophy of "Don't put all your eggs in one basket" to each SKU. At the AWE held in mid - March, Thunderbird showcased its latest product matrix, including:

Thunderbird V3 AI Shooting Glasses (AI glasses without a display): Featuring a high - quality camera and headphones, equipped with the Qualcomm Snapdragon AR1 flagship chip, customized speakers, and Alibaba's exclusive customized Tongyi model, with an AI response speed of 1.3 seconds.

Thunderbird Air3 (Lightweight AR Audio - Visual Glasses): Equipped with the fifth - generation Micro - OLED micro - display, weighing only 76 grams, and for the first time bringing the price of new AR glasses below 2000 yuan.

Thunderbird X3 Pro (AI glasses with a display): Also equipped with Alibaba's exclusive customized multi - modal large - scale model and the Qualcomm Snapdragon AR1 flagship chip; it is also one of the world's lightest mass - producible full - color MicroLED waveguide AR glasses. It uses the first mass - producible surface relief grating etched waveguide technology, reducing the rainbow pattern phenomenon by 95%. It achieves a peak brightness of up to 5000 nits, allowing it to be used outdoors in direct sunlight. It covers functions such as real - time voice and image translation, object recognition, and first - person video creation.

Illustration: From top to bottom are Thunderbird V3, X3 Pro, and Air 3 smart glasses.

This product matrix actually aligns with the technological view that Li Hongwei often mentions: "Climb Mount Everest while laying eggs along the way."

This means that Thunderbird Innovation should have SKUs that use mature technologies, such as the Air series using the BirdBath solution, as well as cutting - edge products representing future trends, such as the X3 series, the first mass - producible full - color MicroLED waveguide AR glasses.

Stratifying product definitions according to technological maturity is also related to Li Hongwei's view on the PMF (Product - Market Fit) stage of the AI/AR category.

Li Hongwei divides the development direction of AI + AR into three stages:

First, AI glasses without a display. Such as the Ray - ban Meta, which has sold millions of units globally.

Second, AR glasses with a display. For example, the AR products for audio - visual viewing under manufacturers like Thunderbird Innovation and Xreal.

Third, AI + AR glasses. That is, the Thunderbird X3 Pro series, which integrates AR display and AI interaction.

Li Hongwei shared with Intelligent Emergence that the first two categories have currently entered the PMF stage in terms of indicators such as sales volume, user attributes (whether they are mass users), and user stickiness.

He mentioned a key indicator: More than 85% of new buyers of AR products for viewing are mass users rather than "enthusiasts."

Then, as the integration of "AI + AR" deepens, is there still a need for these two categories to exist separately?

Li Hongwei gave a direct judgment: "AI glasses without a display are the first stage in the development of AI + AR glasses, and they will inevitably transition to AI glasses with a color display. The latter will definitely replace the former."

Li Hongwei, founder and CEO of Thunderbird Innovation

Here is the edited transcript of the conversation between Intelligent Emergence and Li Hongwei, founder and CEO of Thunderbird Innovation:
 

Intelligent Emergence: Regarding the technological breakthroughs in AR glasses, you mentioned the concept of "Climb Mount Everest while laying eggs along the way." Could you elaborate on this in the context of Thunderbird's technological roadmap?

Li Hongwei: First, what kind of products are those for "climbing Mount Everest"? They are products that view the present from a future perspective.

Currently, many products on the market are monochromatic, and the functions of monochromatic products are relatively limited. The future AI + AR glasses will definitely be all - color, MicroLED waveguide AR all - in - one glasses, which use eye - tracking, gestures, and AI for human - machine interaction.

In this regard, the Thunderbird X3 Pro is such a product for "climbing Mount Everest."

The Air series, which uses the relatively mature BirdBath optical technology, is a product for "laying eggs along the way."

For example, waveguide products without a display/with a monochromatic display and AI shooting glasses are all products for "laying eggs along the way" in our view. They are relatively mature now, and the wearing experience is quite good.

Intelligent Emergence: From the perspectives of software and hardware, which part of Thunderbird's technology stack do you think will face more challenges?

Li Hongwei: Both software and hardware have challenges. The challenges in software mainly lie in the AI part and the core algorithms. We have a lot of work to do in SLAM algorithms, on AR glasses, and in the reasoning of multi - modal models.

I have a background in Internet applications. Its development is in a spiral - upward pattern. Simply put, both software and hardware need to make breakthroughs. If we have to distinguish, hardware breakthroughs are indeed needed first.

How did the iPhone make a breakthrough? It first had a hardware breakthrough. After the touchscreen was introduced, things became relatively simple. The first - generation iPhone already had an APP Store, and its applications were mainly browsers, email, phone calls, and music playback.

Thunderbird is a company with both software and hardware capabilities in this industry. We have more software engineers than hardware engineers, and most of our employees have a background in Internet software. We have made significant investments in software. Recently, an application we developed for the Vision Pro, called Digital Era: The Legend of Elu, is the top - searched game in the Chinese market.

Now, we believe that more investment should also be made in hardware. If a company only has software capabilities without hardware capabilities, we think it is unqualified as a complete device manufacturer.

Intelligent Emergence: For the hardware industry, whether it is AI glasses or AR glasses, they represent a new form of terminal interaction. In this new form, is a new OS system needed? How much investment are we making in the OS direction?

Li Hongwei: A new OS is definitely needed. The most important aspect of an OS is the ecosystem, not just a technical issue. An OS is essentially an abstraction layer of hardware and a bridge between applications and hardware, and this bridge is quite thick.

What kind of OS can be successful? It is an OS that can be used by as many application developers as possible and can be compatible with more hardware. This is a "chicken - and - egg" problem. Only an OS supported by developers and hardware can succeed.

Thunderbird will not develop an OS on its own. This needs to be done by large companies. For example, some large overseas companies are already developing new AI and AR operating systems based on Android.

What Thunderbird is doing is focusing on the key parts of AI + AR glasses based on open - source systems such as Android. For example, we will package AI and AR capabilities into SDKs and provide them for developers to use.

Essentially, we will not pursue developing a new operating system. This is the wrong direction.

Intelligent Emergence: How do you view the relationship between AR glasses and AI glasses? What are the respective positions of these two products in Thunderbird's business map?

Li Hongwei: AI glasses are a very important stage in the industry's development, and this is our positioning for them. Just as the development of mobile phones has gone through the stages of the brick phone, PHS, feature phone, and smartphone, each stage is important and an inevitable part of the development process.

We believe that AI glasses without a display are the first stage in the development of AI + AR glasses, and they will inevitably transition to AI glasses with a color display.

Thunderbird is more advanced in optical display, and optical display technology is essential in the development of glasses. Vision is the most efficient form of human - machine interaction, and users always hope to see images on their glasses. Therefore, we have made important arrangements for AR glasses without a display and are accelerating the layout of AR glasses with a display.

Intelligent Emergence: Your judgment implies another view. If AI evolves to a form with a display, there will be a very obvious integration with AR.

Li Hongwei: Yes. I believe that AI glasses with a display will replace AI glasses without a display.

Intelligent Emergence: How much of your energy is currently spent on AI glasses?

Li Hongwei: We consider two dimensions. The first is the product dimension, which refers to the individual SKUs that users purchase. The second is the internal technological layout dimension, which we analyze separately.

We have AI capabilities, AR capabilities, and capabilities for both display and non - display. This is actually the concept of a technology platform. When developing products, we match the corresponding technologies from this platform.

From the perspective of technological layout, I focus all my energy on AI + AR glasses, and a large amount of investment is made in AI because AI and AR are equally important.

If we focus on products, the corresponding aspects are marketing, product design, and production. So, I spend more of my energy on the underlying aspects of the products.

Intelligent Emergence: So, in your plan, you don't regard AI glasses and AR glasses as two completely unrelated business plans?

Li Hongwei: That's right. In our view, they are like flowers growing from the same root.

Intelligent Emergence: You mentioned the cooperation with Alibaba's Tongyi large - scale model. What are the key focuses of the cooperation?

Li Hongwei: There are two major directions. The first direction is to improve the basic user experience for the form of glasses. For example, the network transmission of large - scale models usually uses the WebRTC protocol. This protocol is very stable under weak network conditions, but the connection establishment speed is relatively slow.

This is not a problem on mobile phones, but on AR glasses, a faster response speed is required because the glasses are on your head, and waiting for two or three seconds for a response would seem slow. So, we hope to make the response speed faster.

The second direction is to modify and upgrade the large - scale model. Since AR glasses can not only have conversations with you but also see what you see and hear what you hear. We hope they can observe 24/7, and not only observe but also perform reasoning and planning. To achieve this, a multi - modal large - scale model is needed.

For example, when you are at Seoul Airport and want to go shopping in a certain place in Seoul and wonder how to get there, the AR glasses will see the airport bus counter in front of you and tell you, "Go there to buy a ticket and take the bus. Get off at the third stop, and then take a taxi for about 20 yuan to reach your destination."

This requires the AR glasses to first understand what you mean, make a plan, adapt to the situation in front of you, and understand that you don't want to spend too much money or go through too much trouble.

Therefore, there is still work to be done in the reasoning and planning of multi - modal large - scale models, and there are many interesting things in this area. Another example is the Agent (AI intelligent agent). How to transform it from a passive to an active form also requires the support of a large - scale model.

Intelligent Emergence: Regarding the sales prospects of AR + AI glasses, do you think it will be a slow - climbing process or an exponential growth?

Li Hongwei: I definitely think it will be exponential growth. It can be divided into several stages.

For example, AI glasses without a display, AI glasses with a display, and AI + AR glasses have different characteristics. So, you will see growth of 10 times, then 3 times, then 10 times again, and then 3 times.

Intelligent Emergence: What stage are we at now?

Li Hongwei: We have passed the first PMF critical point, although it is not the purest "iPhone moment." Recently, almost every time I take a flight, I can see someone wearing Thunderbird products - its first explosion period has begun.

The first PMF I mentioned refers to two things. One is AI glasses without a display. The United States has entered the PMF stage, and China will also reach this stage soon, I estimate. Globally, we believe it has reached the PMF stage.

Second, Thunderbird's Air series, the AR glasses for viewing, has reached the PMF stage. More than 85% of its new buyers are mass users rather than "enthusiasts," and the NPS (Net Promoter Score) is over 45%. That product is also growing rapidly.

Intelligent Emergence: What are the indicators you use to judge whether a product has reached the PMF stage?

Li Hongwei: First, I look at the type of new users, whether they are enthusiasts or mass - market users. We aim for mass - market users who find the product useful rather than just being attracted by the technology.

Second, I check if there is a "word - of - mouth" phenomenon, rather than relying solely on marketing to push the product to users.

Third, I look at user stickiness, that is, whether users use the product frequently after purchasing it.

Of course, these indicators are evaluated based on a certain level of sales volume.