As the mobile computing market approaches saturation, an increasing number of high-tech companies are investing in the research and development of the next generation of computing platforms that have the potential to replace smartphones. To replace smartphones, such a computing platform must first be capable of all-day use, thereby achieving the same or even more daily usage time as smartphones; on the other hand, such a computing platform needs to provide a stronger sense of immersion than smartphones, while also avoiding the inconvenience of having to take smartphones out of one's pocket to use them, thus providing users with more real-time information input and feedback. From this perspective, Augmented Reality (AR) glasses are indeed a very promising computing platform, and as a result, tech giants around the world (including Google, Meta, Huawei, Xiaomi, etc.) are heavily investing in research and development.

The main feature of AR glasses is the use of glasses to provide a device that can interact with users anytime, anywhere, and all day long, and this device can provide important value to users. We believe that AR glasses can be divided into two generations of products. First, let's take a look at the benchmark of AR glasses, that is, what the second-generation AR glasses are like. The most critical feature of the second-generation AR glasses is the ability to superimpose virtual objects on the real world. When users wear the glasses, these virtual objects look no different from real objects. For example, users can place a virtual globe on their desk, and when users slowly move away from the desk, they will see that the virtual globe will still stay on the desk, and it will look smaller in a real way, and when users turn around, the virtual globe will disappear, and only when users turn back to the original angle will the globe reappear, etc. - this technology is called world-locked rendering (WLR). To achieve such WLR capabilities, the second-generation AR glasses require chips that provide strong rendering capabilities and real-time user location tracking capabilities, so as to truly mix virtual objects with the real world.

Advertisement

At present, Microsoft's Hololens is a typical second-generation AR glasses. However, due to the maturity of the technology, the second-generation AR glasses often require a larger size and a high cost, and there is still a long way to go before they can truly become a popular computing platform. Before the second-generation AR glasses can truly be widely deployed, the technically simpler first-generation AR glasses can be mass-produced and deployed on a large scale more simply, thus playing a bridging role. Although the WLR of the second-generation AR glasses can be called revolutionary technology, in addition to the immature WLR technology, AR glasses can also provide other important values to users. The first-generation AR glasses can also be called "smart glasses," whose main feature is that they do not contain the technically expensive WLR technology, but they can provide a series of other important uses, including real-time display of important information, quick shooting, direct calls, etc. At present, companies such as Meta and Huawei have launched the first-generation AR glasses. We expect that in the next five years, the first-generation AR glasses will further expand the market, and the second-generation AR glasses will further mature in technology. Eventually, in the next five to ten years, the second-generation AR glasses will gradually replace the first-generation AR glasses and become the mainstream product after the technology matures and the cost is reduced, becoming a complementary electronic product to smartphones, and even gradually replacing smartphones as the user's preferred computing platform.

Qualcomm's AR2 platform is expected to accelerate the maturity of AR glasses

As mentioned earlier, it seems that the first-generation AR glasses are accelerating the expansion of the market, while the second-generation AR glasses are still working on technology. For AR glasses, the chip plays an extremely important role, and the chip design of AR glasses is full of challenges. The most critical indicators of the AR glasses chip mainly include power consumption, computing power, size, etc.

Firstly, the chip of the AR glasses needs to meet the requirements of low power consumption and high energy efficiency ratio. Due to the design size of the AR glasses, they cannot accommodate a large-capacity battery, and on the other hand, the user experience of the AR glasses is to be able to use it all day long, which puts forward a high requirement for the energy efficiency ratio of the AR glasses chip, that is, to be able to complete complex calculations while using as little energy as possible (to ensure that the battery can support a whole day). In addition to the energy efficiency ratio, it is also necessary to pay attention to the limited heat dissipation capacity of the AR glasses due to its size, so the chip must also be optimized in terms of power consumption to ensure that the glasses do not overheat.

In addition to power consumption and energy efficiency ratio, the processing power of the AR glasses chip cannot be too poor, or the signals that the AR glasses need to process are not necessarily simpler than those of smartphones, and their systems are relatively complex. The WLR required by the second-generation AR glasses is naturally very complex, but even without considering WLR, the first-generation AR glasses still require complex processing capabilities to meet user needs. These capabilities range from the most basic operation of the operating system (commonly modified AR glasses operating systems based on Android systems), high-performance shooting, voice recognition, screen display, running AI models to identify objects in the user's field of view, to more complex hand tracking, eye tracking, etc., all of which require the chip to have the corresponding capabilities and be able to implement these capabilities efficiently.

Finally, due to the size requirements of the AR glasses, the corresponding chip size cannot be too large, which also limits the number of transistors and the process node used (it is necessary to use a process with a smaller feature size to ensure that the chip size is not too large, and processes with smaller feature sizes usually also bring better energy efficiency ratios).

Last week, Qualcomm released the latest AR2 Gen1 series of AR chipsets. We believe that this product is expected to become the preferred chip for most AR glasses in the next few years (or there are not many other choices on the market), and the high performance of this chipset will also promote the further improvement of the performance of AR glasses. For this chipset, the most interesting point is that it includes three chips, including a main chip (AR Processor), a co-processor (AR CoProcessor), and a wireless connection chip. We believe that the main reason why Qualcomm did not release a SoC but released three chips is that the AR glasses have restrictions on size and power consumption. If too many functions are integrated into a SoC, the chip size may be too large, and if multiple modules work at the same time, the power consumption will also be too high, so it is better to split it into three chips to better meet the power consumption and size restrictions.

Among these three chips, the AR Processor is responsible for the most critical work of the AR glasses, including running the SoC and most of the functional modules, including the CPU for running the operating system, the ISP for shooting, the Adreno GPU module for screen display, the Hexagon Processor for accelerating the operation of AI models, etc. As analyzed earlier, to meet the power consumption and size requirements, a 4nm feature size process is used. The AR CoProcessor is mainly responsible for eye tracking and related display optimization (forveated rendering), and the last wireless interconnection chip is responsible for high-speed interconnection with the mobile phone, supporting WiFi7.Overall, Qualcomm's AR2 Gen1 has achieved quite good performance, with its AI-related performance being 2.5 times that of the previous generation product used in VR (XR2), while power consumption is half of it. Leveraging Qualcomm's leadership in the wireless field, Qualcomm also emphasized the importance of using wireless interconnect chips to work in conjunction with mobile phones. In Qualcomm's vision, the most critical aspect of AR glasses chips is the ability to perform some real-time processing (including the operation of some basic AI models), while more complex operations, such as WLR or larger models, can be offloaded to the mobile phone by using wireless interconnect methods to transmit data for execution. We believe that in the first generation of AR glasses, Qualcomm's AR Processor chip is the most promising to be widely adopted, as it indeed includes most of the functional modules needed by AR glasses, while the AR CoProcessor mainly targeting eye tracking may not be widely applied at the beginning because eye tracking is not a standard feature in the first generation of AR glasses. Finally, whether the first generation of AR glasses needs to maintain a long-term high-speed wireless connection with the mobile phone is also inconclusive (currently, many operations can be directly completed on the glasses, and only a few operations, such as information retrieval that requires connection to the internet, need to be connected to the mobile phone). Therefore, whether AR glasses system vendors will use the wireless solution in Qualcomm's AR2 Gen1 or use the previous generation of wireless interconnect solutions from Qualcomm or other manufacturers to reduce costs still needs to be tested by time.

How will the future AR chip market competition landscape be?

As the first generation of AR glasses market continues to expand, we believe that Qualcomm's AR2 Gen1 is currently in a leading position in terms of related solutions, because Qualcomm is willing to invest when the market capacity of AR glasses has not yet been fully developed. However, as the market capacity of AR glasses increases year by year, we expect that other chip companies such as MediaTek will also join the competition. From a technical perspective, in addition to the need for high energy efficiency and low power consumption design, most of the design in Qualcomm's AR2 Gen1 can also be achieved by competitors such as MediaTek. Therefore, it is also fully expected that if the market is large enough, competitive companies will also join the market.

Beyond the first generation of AR glasses, the second generation of AR glasses is currently another market for Qualcomm. The main challenge for third-party manufacturers like Qualcomm in the second generation of AR glasses is that system manufacturers (such as Apple, etc.) will be more inclined to use their own chip solutions. On one hand, this is because the second generation of AR requires more extreme performance and the market is smaller, making it difficult for third-party solutions to make very detailed designs for such a small market. On the other hand, using self-developed chips can strengthen competitive barriers, thereby ensuring that their products have unique technical performance that cannot be achieved by other competitors simply by purchasing the same third-party chipsets.