Core Insights - Wallen Technology is a leading domestic AI chip company focusing on GPGPU technology and related products, with a strong emphasis on innovation and market opportunities [1][2] Group 1: Company Overview - Founded in 2019, Wallen Technology has a diverse team with backgrounds from major companies like AMD and Huawei, and holds 1,158 global invention patents as of June 2025, ranking first among Chinese GPGPU companies [1] - The company primarily offers GPGPU, boards, modules, servers, and cluster products, projecting revenues of 336.8 million yuan in 2024 and 58.9 million yuan in the first half of 2025, indicating steady commercialization progress [1] Group 2: Product and Technology - Wallen Technology focuses on GPGPU architecture and is pioneering in Chiplet technology and optical switching, with the BR166 achieving double the computing and memory performance of the BR106, and a bidirectional bandwidth of 896 GB/s [2] - The hardware supports various deployment forms like OAM and PCIe, while the software platform is compatible with mainstream frameworks like PyTorch, showcasing the company's full-stack optimization capabilities [2] Group 3: Financial Performance - The company's gross margin is projected to be 76.4% in 2023, decreasing to 53.2% in 2024, and further to 31.9% in the first half of 2025 due to an increase in entry-level product sales [2] - Customer diversification is improving, with the top five customers accounting for 97.9% of revenue in the first half of 2025, and the largest customer contributing only 33.3% of total revenue, indicating reduced dependency on major clients [2] Group 4: Ecosystem and Supply Chain - Wallen Technology collaborates with companies like Xizhi Technology and ZTE to launch the LightSphere X optical switching super node, which supports large-scale deployments and adapts to the needs of large model training [3] - The company has achieved self-sufficiency in its supply chain, completing domestic replacements for key components, and plans to commercialize the next-generation BR20X architecture by 2026 without being affected by the BIS entity list [3] Group 5: Investment Outlook - Wallen Technology is rated as a "Buy," with projected revenues of 950 million yuan, 2.02 billion yuan, and 3.95 billion yuan for 2025-2027, and adjusted net profits of -830 million yuan, -630 million yuan, and 74 million yuan respectively [4] - The current market valuation corresponds to a price-to-sales ratio of 20x for 2027E, with a weighted average of 28x for comparable companies, reflecting the significant potential in domestic AI capital expenditures [4]

Investment Rating - The report assigns a "Buy" rating for the company, Wallrun Technology (壁仞科技) [2][7]. Core Insights - Wallrun Technology is a leading domestic AI chip company focusing on GPGPU architecture and intelligent computing solutions, with a strong emphasis on proprietary technology and a diverse team background [6][15]. - The company has achieved significant revenue growth projections, with expected revenues of RMB 9.5 billion, RMB 20.2 billion, and RMB 39.5 billion for the years 2025 to 2027, respectively [7]. - The report highlights the company's innovative product offerings, including the BR106 and BR166 chips, which are designed for large-scale AI training and inference applications [25][28]. Financial Data and Profit Forecast - Revenue projections for Wallrun Technology are as follows: RMB 62 million in 2023, RMB 337 million in 2024, RMB 945 million in 2025, RMB 2,021 million in 2026, and RMB 3,951 million in 2027, with year-on-year growth rates of 12,330.86%, 442.97%, 180.61%, 113.83%, and 95.51% respectively [5]. - Adjusted net profit forecasts indicate losses of RMB 1,051 million in 2023, RMB 767 million in 2024, RMB 827 million in 2025, and a reduced loss of RMB 632 million in 2026, with a projected profit of RMB 74 million in 2027 [5][7]. - The company's gross margin is expected to fluctuate, with rates of 76.4% in 2023, 53.2% in 2024, and 31.9% in the first half of 2025, primarily due to changes in product sales mix [6][35]. Technology and Product Development - Wallrun Technology focuses on GPGPU architecture and has developed a comprehensive hardware and software ecosystem, including the BIRENSUPA software platform, which supports major AI frameworks [20][31]. - The company is pioneering advanced technologies such as Chiplet architecture and optical interconnects, enhancing the performance and scalability of its AI computing systems [50][53]. - The BR20X chip is expected to be commercialized in 2026, featuring improved performance and support for various data formats, further solidifying the company's market position [29][30]. Market Position and Ecosystem - Wallrun Technology has established strong partnerships with major telecommunications operators and is expanding its customer base, with a projected revenue contribution from its top five customers decreasing over time [6][38]. - The report emphasizes the growing domestic AI capital expenditure (Capex) market, which is expected to accommodate multiple AI chip companies, indicating a favorable environment for Wallrun Technology's growth [8][7]. - The company has successfully implemented a domestic supply chain strategy, ensuring production and research continuity despite external challenges [54].

Core Viewpoint - The article discusses the future device technologies for optical circuit switches (OCS), focusing on their application in data center networks and machine learning supercomputers, highlighting key performance parameters that affect system performance and reliability [2][4]. Group 1: Introduction and Background - Large-scale systems rely on networks to transmit information from source to destination, primarily using electrical packet switches (EPS) and a fixed Clos topology, which face scalability limitations in cost, latency, and reconfigurability [4]. - The exploration of optical circuit switches (OCS) aims to dynamically adjust network topology to match communication patterns, leading to their deployment in large-scale data centers and machine learning systems [4][6]. Group 2: Future Optical Switching Technologies - Table I outlines key performance metrics for various commercial and developmental OCS technologies, including port count, switching time, insertion loss, and driving voltage, which vary based on whether the switching function is implemented in free space or guided-wave structures [8][9]. - Current commercial OCS devices are based on customized hardware and control schemes, with no single switch technology achieving optimal performance across all applications [10]. - MEMS-based optical switches provide significant cost advantages in large-scale data center networks and enhance system performance when used in TPU superpods [13]. Group 3: Device Technologies - Emerging device technologies include non-mechanical two-dimensional digital liquid crystal (DLC) pixel arrays, which control light beam propagation direction using polarization characteristics [13]. - Two-dimensional devices are primarily based on cross-matrix structures with waveguides, with silicon photonics (SiP) technology being a focus for achieving lower costs and faster switching speeds [16]. - Challenges for two-dimensional switches include high losses during fiber coupling and limited port counts, with interference-based devices and heterogeneous integrated devices being explored to address these issues [16][17]. Group 4: Conclusion - As optical circuit switching technology commercializes, research activities around future optical switch device technologies are rapidly increasing, with expectations for some developmental technologies to be introduced into future computing and networking systems for mass production [22].

Core Viewpoint - Google discusses the future device technology for optical circuit switches (OCS), focusing on data center networks and machine learning supercomputers, highlighting the impact of device parameters on system performance and reliability [1][3]. Group 1: Introduction and Background - Large-scale systems rely on networks to transmit information from source to destination, primarily using electronic packet switches (EPS) and a fixed Clos topology, which face scalability limitations in cost, latency, and reconfigurability [3][5]. - Early research into OCS aimed to dynamically adjust network topology to match communication patterns, leading to practical deployments in large-scale data centers and machine learning systems [3][4]. Group 2: Future Optical Switching Technology - Table I outlines key performance metrics for commercial and developmental OCS technologies, including port count, switching time, insertion loss, and driving voltage, with variations based on whether the switching function is implemented in free space or guided-wave systems [7][8]. - Current commercial OCS devices are based on customized hardware and control schemes, with no single technology achieving optimal performance across all applications [9]. Group 3: Existing and Emerging Technologies - MEMS-based optical switches provide significant cost advantages in large-scale data center networks, enhancing system availability and performance [12]. - New device types, such as two-dimensional digital liquid crystal (DLC) pixel arrays, utilize polarization properties for light beam direction control, allowing for scalable port configurations [12][14]. - Development of two-dimensional devices primarily focuses on silicon photonics (SiP) technology, which is compatible with standard CMOS processes, aiming for lower costs and faster switching speeds [14][15]. Group 4: Challenges and Innovations - The Apollo project by Google aims to replace traditional network infrastructure with OCS, reducing communication delays and power consumption by keeping data in the optical domain [25][29]. - The OCS technology allows for significant reductions in power consumption, with maximum power usage at 108 watts compared to 3000 watts for traditional EPS systems [30]. - Google has deployed thousands of OCS systems, claiming it to be the largest application of OCS globally, with substantial cost and energy savings [31]. Group 5: Future Directions - Google is focused on developing OCS systems with higher port counts, lower insertion losses, and faster reconfiguration speeds, which are expected to enhance efficiency and reliability [36][38]. - The company believes that modern data centers can achieve bandwidth comparable to the entire internet, indicating the potential for massive communication volumes [38].

Core Insights - The report from Huazhong Securities highlights the increasing demand for Optical Circuit Switches (OCS) due to the high requirements of AI large model training for communication bandwidth, latency, and power consumption [1][3] - The global OCS market is projected to grow from $0.07 billion in 2020 to $0.78 billion by 2025, with a compound annual growth rate (CAGR) of 62% [3] - The OCS industry chain consists of upstream core components, midstream equipment integration, and downstream applications, with high technical barriers and significant value concentration in upstream components [4] Market Growth - The OCS market is expected to reach $2.02 billion by 2031, with a CAGR of approximately 17.2% from 2025 to 2031 [3] - Major players in the market include Google and Coherent, with the top four companies expected to hold about 69% of the market share by 2025 [3] Applications and Efficiency - OCS is primarily applied in three scenarios within AI computing clusters: Scale-Up, Scale-Out, and Scale-Across, showcasing its efficiency in expanding TPU cluster capabilities [2] - For instance, a Google TPU cluster with 4096 TPU v4 chips requires 48 OCS switches, demonstrating a TPU to OCS ratio of 85:1, which improves to 192:1 with future upgrades [2] Industry Participants - Companies like Yintan Zhikong and Saiwei Electronics are positioned to benefit from the OCS market growth, with Yintan focusing on semiconductor design and manufacturing, and Saiwei leading in MEMS technology [5] - Yintan plans to acquire Guilin Guanglong Integration to enhance its OCS capabilities, while Saiwei is set to start mass production of MEMS-OCS, capitalizing on the expanding AI computing demand [5]

Core Insights - The OCS industry is currently dominated by overseas companies, but domestic manufacturers are gradually increasing their participation, with a focus on component supply or OEM opportunities [1] - The OCS market is expected to experience significant growth, with projections indicating a market size of approximately $400 million in 2025 and over $1.6 billion by 2029, reflecting a CAGR of about 41% [2] - Four main technological routes (MEMS, liquid crystal, piezoelectric, and silicon photonic waveguides) are evolving together, with no single solution currently holding a dominant position [3] Investment Directions - Key investment directions include manufacturers already involved in overseas OCS supply/OEM, those with potential to enter the overseas market, and optical module manufacturers actively engaging in OCS [1] - The establishment of the OCS sub-project by OCP is expected to promote industry standardization, while domestic operators and the Ministry of Industry and Information Technology are pushing for OCS application implementation [2] Technical Considerations - OCS technology is particularly advantageous for data centers, enhancing overall network performance, operational efficiency, and sustainability [1] - The coexistence of CPO and OCS in data center networks is anticipated, with CPO being more suitable for dynamic traffic scenarios and OCS for stable traffic patterns [4] - The application of OCS is currently advancing faster than that of CPO, particularly in the Spine layer of scale-out networks [4]

Core Insights - Google is making a significant investment in optical circuit switches (OCS), with an expected procurement of over 23,000 units by 2025, driven by the demand for its TPU chip clusters [1] - The global OCS market is projected to grow from $36.6 million in 2024 to $2.022 billion by 2031, reflecting a compound annual growth rate (CAGR) of 17.1% [5] - OCS technology is becoming essential for AI computing clusters, as it meets the high bandwidth and low latency requirements necessary for AI model training [11] Investment and Market Dynamics - Google's investment in OCS systems is expected to reach $3 billion by 2026, with a significant order of 4,600 units from Polatis valued at $250 million [3] - The OCS market is experiencing rapid growth, with a CAGR of 49.8% from $7.278 million in 2020 to $36.6 million in 2024, and is anticipated to exceed $2 billion by 2031 [12] Technological Advancements - OCS technology offers advantages over traditional electrical switches, including a theoretical efficiency that can be 1,000 times greater and a power consumption reduction of around 40% [6] - Three main technological pathways exist in the OCS field: MEMS, DLC, and DLBS, with DLBS (piezoelectric ceramics) showing the most promise for high stability AI cluster interconnections [7][10] Competitive Landscape - The OCS market is currently dominated by North American companies, which hold 58% of the market share, while Chinese manufacturers are rapidly gaining ground with a 28% share of global production capacity by 2024 [13] - Chinese companies like Dekoli and Guangxun Technology are innovating in the OCS space, with Dekoli's photon routing engine achieving latency below 10 microseconds and Guangxun being the only domestic producer of MEMS optical switches [13][14] Future Outlook - The demand for OCS is expected to surge as AI server interconnect needs increase, with estimates suggesting that 48 OCS switches are required for every 4,096 TPU units [11] - The collective actions of major tech companies, including Google, Meta, and Nvidia, are accelerating the adoption of OCS technology, marking a significant shift in data center infrastructure [16]

Summary of OCS Switch Technology, Trends, and Market Potential Industry Overview - The OCS (Optical Circuit Switch) technology is applied in traditional and AI data centers, enabling low-power and efficient data transmission through optical channels. The market potential is projected to be 30-40 times larger than the current electrical switch market [1][6][26]. Key Insights and Arguments - **Market Growth**: Google has scaled up the deployment of OCS in its TPU network, purchasing approximately 12,000 units. If other potential clients like Microsoft, NVIDIA, Oracle, and Meta adopt OCS at scale, the market could expand significantly, with expectations of reaching 300,000 units in the coming years [1][6][26]. - **Technological Advancements**: Google’s OCS switch has achieved a thousand-fold increase in speed while reducing power consumption, showcasing significant progress in the switch technology sector [2][4]. - **Core Advantages**: OCS technology eliminates the need for optical-electrical conversion, directly interconnecting different cards through optical channels, which significantly reduces power consumption. For instance, a 51.2T electrical switch consumes about 3 kW, while OCS can save nearly half of that power [1][9]. - **Industry Initiatives**: The establishment of the Optical Switch Project Group by the Open Compute Project (OCP) indicates a growing recognition of OCS as a complement to traditional electrical switches and CPU switches [1][7]. Important Developments - **Domestic Collaborations**: In China, Xizhi Technology, in collaboration with ZTE and Biran Technology, launched a full-optical OCS supernode network, indicating a trend towards advanced optical interconnection solutions [1][8]. - **Application in AI Networks**: OCS is seen as a future trend in AI networks due to its low power and low latency characteristics, although current switching times may still favor electrical switches for frequent data exchanges [13][15]. - **Investment Returns**: NVIDIA has reported that by adding OCS devices, they can achieve automatic switching in case of hardware failures, reducing the risk of interruptions in large model training, with an investment payback period of about one to two years [4][14][16]. Market Dynamics - **Future Demand**: Google is expected to procure between 16,000 to 20,000 OCS switches by 2026, with a potential market value of around $1 billion based on a unit price of $50,000 [25]. - **Long-term Market Potential**: The OCS market is projected to exceed $2 billion by 2029, with significant interest from major players like Microsoft, Oracle, and NVIDIA, indicating a shift towards OCS solutions [26][30]. - **Competitive Landscape**: Traditional electrical switches currently dominate the market, with an estimated 800,000 to 1 million units. However, OCS is expected to capture a 30% market share in the next few years, potentially reaching over 300,000 units by 2030 [30]. Additional Insights - **Technological Variants**: OCS technology includes various solutions such as MEMS, silicon-based liquid crystal, and piezoelectric ceramics, each with unique advantages and challenges [5][28]. - **Supply Chain Dynamics**: Key suppliers for Google’s OCS switches include Lumentum, Claudeline, and Coherent, with significant contributions from domestic manufacturers like Saiwei Electronics and Guanghua Technology [19][32][33]. - **Adoption Trends**: While overseas data centers like Google have adopted OCS extensively, domestic centers are beginning to explore its use, indicating a gradual shift in the industry [29]. This comprehensive overview highlights the transformative potential of OCS technology in the data center industry, emphasizing its advantages, market dynamics, and future growth prospects.

Core Viewpoint - Optical Circuit Switches (OCS) differ from traditional switches by enabling direct optical switching without optical-electrical conversion, leading to reduced hardware costs and power consumption [1][2][5]. Group 1: Overview of Optical Switches - OCS operates by forwarding optical signals through wavelength division multiplexing and cross-connection, with optical-electrical conversion occurring at the server side [1][2]. - The main performance indicators of optical switches include switching speed, insertion loss, return loss, durability, and reliability [2][3]. Group 2: Current Development Status - OCS technology has gained attention due to Google's initiatives, with increasing applications in data centers [5][6]. - The rise of east-west traffic in large data centers has created bandwidth bottlenecks in traditional three-layer architectures, necessitating the adoption of the Spine-and-Leaf architecture for better scalability and reduced latency [6][8]. Group 3: Market Demand and Shipment Volume - Global shipments of optical switches reached 10,000 units in 2023, with an expected increase to approximately 13,000 units in 2024, driven by growing bandwidth demands in machine learning and cloud computing [16][19]. Group 4: R&D Directions - Despite the promising outlook for optical switch technology, challenges remain, including high initial capital expenditures and signal insertion loss [14][19]. - Continuous technological advancements are anticipated to enhance performance and reduce manufacturing and deployment costs, solidifying the foundation for large-scale applications [19].