Listing Success | Frost & Sullivan Supports Beijing Haiguang Chip Technologies Co., Ltd. in Successful Listing on Hong Kong Stock Exchange (1191.HK)

During the Hong Kong listing process, Frost & Sullivan undertook the following tasks: helping the issuer accurately and objectively understand its position in the target market, using objective market data to identify and highlight the issuer’s competitive advantages, assisting the issuer, investment banks, and other intermediaries in preparing important sections of the prospectus such as overview, competitive advantages and strategy, industry overview, and business, facilitating communication between the issuer and the Stock Exchange and investors, helping investors quickly understand the market ecosystem and competitive landscape, and supporting the issuer in responding to various questions from the Stock Exchange regarding the industry.

Frost & Sullivan has always been a leader in helping companies list on the Hong Kong stock market. According to LiveReport Big Data (data as of March 31, 2026), over the past 36 months and 12 months, and from January to March 2026, Frost & Sullivan provided listing industry consulting services for 195 Hong Kong IPOs (market share of 71%), 101 IPOs (market share of 73%), and 30 IPOs (market share of 77%) respectively, ranking first in number of services provided. Frost & Sullivan possesses rich industry experience and communication skills with regulatory authorities, exchanges, investment and financing institutions, and related organizations.

•The company is one of the few enterprises globally that possesses both independent design capabilities for silicon photonics chips and R&D and mass production capabilities for silicon photonics optical modules;

•The company is among the first in China to achieve mass production and delivery of 400G and 800G AI optical modules;

•The company has the ability to develop 3.2T and 6.4T optoelectronic chips, which are applied in NPO and CPO forms based on silicon photons;

•The company’s silicon photonics chips use the “less-change CMOS” design. By adopting this approach, the production cost of the company’s silicon photonics chips can be reduced by 30% to 40% compared to overseas production, and the overall cost of silicon photonics optical modules can be reduced by 20% to 30% compared to competing products.

According to Frost & Sullivan’s report, based on AI optical module revenue in 2025, the company:

•Ranks eighth globally among Chinese optical module suppliers;

•Occupies a 1.6% market share in the global AI optical module market.

In recent years, the rapid development of the AI industry has put higher demands on communication networks, significantly driving innovation and application of optoelectronic interconnection technologies. In terms of data center network architecture, AI workloads are pushing traditional networks toward more intelligent and efficient directions, requiring simultaneous improvements in horizontal and vertical expansion capabilities. Horizontal expansion relies on high-bandwidth lossless interconnection between computing nodes, while vertical expansion focuses on high-speed data interaction at the chip and server levels. This shift has led to new types of data center networks with dynamic sensing and flexible resource scheduling capabilities. Optoelectronic interconnection technology, due to its significant advantages in high-density and low-power interconnection, has become the core foundation for achieving high-speed connections within and between data centers, thereby strongly supporting the coordination and global scheduling of distributed computing resources driven by AI.

By product type, optoelectronic interconnection products mainly include AI pluggable optical modules, co-packaged optics/near-packaged optics (CPO/NPO), and active copper cables (AEC). The pluggable optical module connects to servers, GPUs, or other devices at one end to receive electrical signals, and to optical fibers at the other end, enabling bidirectional conversion between electrical and optical signals, thus completing data transmission between devices. CPO and NPO integrate the optical engine with exchange chips or AI accelerator chips, systematically achieving the highest bandwidth density and lowest transmission energy consumption by shortening electrical interconnection paths. They are the core solution for building next-generation super-large computing clusters. AEC integrates chips at both ends of the copper cable, effectively improving signal transmission quality while maintaining low cost and low power consumption, resulting in better transmission performance than passive copper cables.

Driven by the explosive growth of AI computing demand and large-scale deployment of data centers, sales of optoelectronic interconnection increased rapidly from 2021 to 2025. In the future, as technology continues to evolve to 1.6T/3.2T and new solutions like CPO and NPO are commercialized, this growth trend is expected to continue. The global optoelectronic interconnection market size in 2025 was 7.4 billion yuan, rising to 80.1 billion yuan in 2026, with a compound annual growth rate of 81.2%; it is expected to reach 567.4 billion yuan by 2030, with a compound annual growth rate of 43.9 from 2026 to 2030.

Source: Frost & Sullivan analysis

AI optical modules refer to optical modules used to meet the high bandwidth and low latency requirements of AI computing clusters. These products are typically deployed in AI data centers to address the bottleneck of massive data exchange. In terms of transmission rates, due to the high demands on data throughput and transmission capacity of AI computing clusters, AI optical modules mainly consist of 100G and above optical modules, including 100G, 200G, 400G, 800G, 1.6T, and 3.2T types. Additionally, by technical route, AI optical modules can be divided into silicon photonics optical modules, which have significant cost and power advantages in ultra-high speed scenarios, and non-silicon photonics optical modules.

The optical module industry chain includes three main segments: upstream, centered on material and component supply, including optical chips (including silicon photonics chips), integrated circuit chips, PCBs, and structural components. Midstream is responsible for the integration and manufacturing of optical modules, achieving efficient data transmission through the integration of optoelectronic components. Downstream customers are mainly cloud service providers, who purchase AI optical modules for AI data centers.

AI optical modules are segmented by end-use requirements, and this classification is commonly used in the industry, with two key reasons: (1) Different application scenarios have distinct requirements for product performance. Optical modules used in AI data centers need to meet high bandwidth, low latency, and high-density transmission standards, supporting large-scale parallel data exchange and high-load computing operations; while traditional general-purpose data centers and telecommunications networks prioritize stability, long-distance transmission, and cost control, with relatively relaxed requirements for peak bandwidth and latency. (2) Specific market demands are clear and continuously expanding. Global cloud service providers specifically purchase optical modules suitable for AI computing clusters, making this category an independent high-growth segment in the optical module industry. In 2025, the AI optical module market size accounted for 44% of the total optical module market, with increasing industry importance.

Driven by the rapid development of large language model training, inference, and commercial applications, global cloud service providers and large internet companies have massively invested in building AI data centers. As a key “data artery” for high-speed internal interconnection in data centers, the demand for AI optical modules has surged. From 2021 to 2025, the global AI optical module market size increased from 7 billion yuan to 71.8 billion yuan, with a compound annual growth rate of 79.0%. In the future, with the iteration of next-generation high-speed products such as 1.6T and 3.2T, and the commercialization of low-power new architectures, the market size will continue to expand. It is expected that by 2030, the global AI optical module market size will reach 347.5 billion yuan, with a compound annual growth rate of 30.6 from 2026 to 2030.

Source: Frost & Sullivan analysis

From a technical perspective, silicon photonics is a technology system based on silicon semiconductor materials, which uses CMOS-compatible processes (compatible with traditional chip manufacturing) to integrate functions such as light signal generation, transmission, modulation, and detection onto a single silicon chip. Its core value lies in overcoming the limitations of traditional separate optoelectronic devices (such as high cost, low integration, and high power consumption) by leveraging the cost-effectiveness and mature semiconductor manufacturing processes of silicon. This allows optoelectronic components to be miniaturized and mass-produced, while achieving high speed and low power consumption, making silicon photonics the key bridge between “optical communication” and “semiconductor” technologies.

Driven by the surge in high-speed and high-bandwidth demands due to AI computing growth and the continuous maturity and improvement of silicon photonics technology, the AI optical module market using silicon photonics has experienced rapid growth, from 2 billion yuan in 2021 to approximately 29.9 billion yuan in 2025, with a compound annual growth rate of about 95.5%. Looking ahead, the AI optical module market using silicon photonics is expected to further expand, with expected sales reaching 192.5 billion yuan by 2030, and a compound annual growth rate of 35.8 from 2026 to 2030.

Source: Frost & Sullivan analysis

Benefiting from the rigid demand for high-speed interconnection caused by the soaring demand for AI and computing, and the national “East Data, West Computing” project promoting the integrated layout of big data centers across the country, the Chinese AI optical module market size increased from 1.6 billion yuan in 2021 to 16.7 billion yuan in 2025, with a compound annual growth rate of 79.6%. During the forecast period, the market is expected to grow further to 99 billion yuan by 2030, with a compound annual growth rate of 36.0 from 2026 to 2030.

Technically, similar to global trends, driven by the explosive growth of AI computing demands, accelerated data center upgrades, and significant progress in Chinese silicon photonics technology, the Chinese AI silicon photonics optical module market has experienced rapid growth, from 300 million yuan in 2021 to 6.6 billion yuan in 2025. Looking ahead, with the commercialization of next-generation products such as 1.6T optical modules and continued policy support for the semiconductor and photonics industries in China, the market is expected to continue to grow explosively to 55.9 billion yuan by 2030, with a compound annual growth rate of 45.1 from 2026 to 2030.

Source: Frost & Sullivan analysis

Participants in the AI optical module industry can be divided into two categories: third-party optical module suppliers and in-house optical module manufacturers. Third-party participants are companies that mainly develop and produce AI optical modules for external sales rather than internal use. In contrast, in-house manufacturers produce such modules mainly for internal use.

The total market size of the global AI optical module industry in 2025 was 71.8 billion yuan. In 2025, the company generated 1.1 billion yuan in AI optical module revenue, ranking eighth among Chinese AI optical module suppliers and accounting for 1.6% of the global market share.

Source: Annual Report, Frost & Sullivan analysis

●Explosive Growth of AI Computing Demand

Artificial intelligence, especially generative AI and large model training, has set unprecedented requirements for data transmission rates and bandwidth within data centers. Traditional computing clusters cannot meet the large-scale data exchange generated by thousands of GPUs working together, which directly drives the iteration and scale expansion of high-speed interconnection optical modules. 800G optical modules have become the mainstream choice for current AI training clusters, and higher-rate products such as 1.6T are in the pre-commercial stage. It is expected that as the parameters of AI models continue to grow and application scenarios deepen, the demand for high-speed and low-power optical modules will remain strong and will be the core driver of the market.

●Global Data Center Upgrades and Cloud Service Expansion

Global digital transformation is driving the construction of super-large data centers and the upgrade of existing data center network architectures, evolving from 100G/400G to 800G and above. The popularity of applications such as cloud computing, edge computing, and streaming media and IoT has led to a continuous increase in global data traffic, putting higher bandwidth requirements on internal and inter-data center connections. This trend not only increases the usage of optical modules but also drives technology toward lower power consumption and higher density (such as silicon photonics and CPO). Driven by national projects like “East Data, West Computing” in China, data center construction is accelerating, providing a broad and continuous domestic market for optical modules.

●National Policy Support for Digital Infrastructure

Governments around the world view computing infrastructure and broadband networks as strategic assets and implement policies to support their development. Initiatives such as China’s “East Data, West Computing” project and the EU’s “2030 Digital Compass” aim to enhance high-performance computing and data center capabilities while building high-speed and comprehensive network infrastructure. These policies not only directly stimulate data center construction and high-speed optical communication demand but also provide strong policy and ecosystem support.

●Iteration and Commercialization Breakthroughs in Optical Modules

In recent years, optical modules have made significant breakthroughs in materials, packaging, and integration technologies, effectively supporting the simultaneous achievement of high speed, low power consumption, and cost reduction goals. Advanced solutions represented by silicon photonics have gradually matured and entered the stage of large-scale commercialization, demonstrating excellent performance and integration advantages in high-speed scenarios such as 800G/1.6T. At the same time, new architectures such as CPO and NPO have made substantial progress in reducing system power consumption and latency, providing key technical support for next-generation data centers and AI clusters. The continuous innovation and commercialization of these technologies are expanding the performance boundaries of optical modules and promoting their widespread use in high-speed data communication, thus becoming an important technological driving force for market development.