The popular story of artificial intelligence begins with the processor. Bigger models require more GPUs, more power and ever-larger data centers. Yet once thousands of accelerators are assembled into a single computing cluster, another problem emerges: the chips must constantly exchange enormous volumes of data, and they are only as useful as the network connecting them.
This is where optical transceivers enter the picture.
Inserted into switches and other networking equipment, these compact devices convert electrical signals into light, send them through optical fiber and convert them back again. They rarely appear in photographs of AI supercomputers. But without them, the GPUs at the center of the AI boom would spend more time waiting for data and less time processing it.
That largely invisible layer of infrastructure has propelled Ligent Technologies—known in Chinese as Nazhen Technology—toward one of Hong Kong’s larger technology listings of the year.
The Qingdao-headquartered company is offering approximately 172 million shares at HK$32.96 each, raising about HK$5.67 billion before expenses. Trading is scheduled to begin on September 22 under stock code 9856. The listing would give Ligent an expected market capitalization of approximately HK$32.4 billion. Its Hong Kong prospectus presents the company as a supplier of optical transceivers, optical chips and optical network terminals.
But the more revealing story is not the size of the offering. It is how a company built for an earlier generation of broadband infrastructure repositioned itself inside the supply chain for AI.
Before AI, There Was Broadband
Ligent did not begin as an AI infrastructure company.
Its history reaches back to two businesses established more than two decades ago. Ligent Tech was formed in the United States in 2002 by Chinese electronics group Hisense and TransLight, while Qingdao Broadband was established the following year. Founder Huang Weiping, an optical-communications specialist who spent much of his academic career in Canada, helped connect the company’s Chinese manufacturing base with expertise developed in North America.
The group’s original opportunity was fiber broadband.
During the 2000s and early 2010s, telecom operators around the world were replacing copper connections with fiber. The industry called this transition FTTx—“fiber to the x,” with the final letter representing a home, building or other destination. Every new fiber connection required components capable of turning electronic data into optical signals.
This was a less glamorous market than AI computing, but it was enormous. It gave Ligent experience in producing optoelectronic components at scale, meeting telecom reliability requirements and competing in a market where prices tend to fall as each product generation matures.
According to the prospectus, Ligent reached the world’s largest share of the FTTx optical-transceiver market in 2012. That legacy remains visible in the company today: it still sells transceivers and terminal equipment used in residential broadband, telecom networks and Wi-Fi systems.
Yet Ligent’s management was already looking beyond the last mile.
In 2011, the company acquired a data-communications business from SAE Technologies Development in Dongguan. That transaction took Ligent into the market for components connecting servers, storage systems and data-center switches.
The distinction would become crucial. Telecom networks move information across cities and countries. Data-center networks move immense quantities of information among machines inside computing facilities. AI would eventually turn the second category into the company’s principal engine of growth.
Building From the Chip Outward
Ligent’s next step was to move deeper into the components inside an optical transceiver.
In 2012, it acquired LigentCom, a US manufacturer of optoelectronic components. The deal brought Fabry–Pérot and distributed-feedback laser technologies into the group. A year later, an asset acquisition from US company Multiplex added electro-absorption modulated laser and tunable-laser technologies.
These names describe different methods of producing and controlling the light used to carry information. For customers, however, the strategic point is simpler: Ligent was no longer only assembling the box. It was acquiring the ability to develop some of the laser chips inside it.
That distinction matters because the optical chip helps determine a transceiver’s speed, power consumption, transmission distance and cost. Bringing chip development and module production under one roof can shorten development cycles and give a supplier more control over performance and supply.
Ligent says it is now among a relatively small number of companies capable of both developing and producing optical transceivers and optical chips. Its research network spans Qingdao, Wuhan, Silicon Valley and Singapore, while its patent portfolio comprised 1,581 granted patents and 726 pending applications as of September 5, 2026.
The company has also recruited management from the international optical-communications industry. CEO Hong Jin joined Ligent in 2023 after serving as a vice president in Intel’s data-center business group and general manager of its silicon-photonics product division. He became Ligent’s CEO in May 2025.
The timing of that appointment coincided with a much larger change in the company’s business.
The Pivot Hidden in the Numbers
In 2023, Ligent generated RMB1.06 billion from data-communications transceivers. These products accounted for about a quarter of its revenue.
By 2025, the same business was producing RMB5.47 billion and represented 65.5% of total revenue. During the first half of 2026, its share rose again to 69.4%.
In just three years, datacom transceivers had gone from one part of a diversified optical-communications company to the business defining it.
Ligent’s total revenue increased from RMB4.24 billion in 2023 to RMB8.35 billion in 2025. Revenue reached RMB5.39 billion in the first six months of 2026, up 27.9% from the corresponding period a year earlier. First-half profit increased 29.7% to RMB661 million.
The company attributes much of that acceleration to demand from AI and cloud-computing customers. Its datacom transceiver revenue rose from RMB2.74 billion in the first half of 2025 to RMB3.74 billion one year later. Over the same period, its overall gross margin increased from 18.7% to 24.2%, supported by a greater contribution from higher-margin products and overseas markets. The prospectus provides the full product and financial breakdown.
This is the financial signature of the AI infrastructure boom. More accelerators require faster links between servers, switches and racks. As clusters grow, the number of connections can rise rapidly, while each new generation must carry more data without consuming proportionally more electricity.
Ligent says it has developed and entered mass production of 800-gigabit and 1.6-terabit optical transceivers. It is also researching 3.2T transceivers and 6.4T optical engines, and has submitted samples of a 3.2T near-packaged optics product for customer validation.
The numbers refer to the amount of data a module can transmit each second. They are also markers in a relentless product cycle: 400G gives way to 800G, then 1.6T and eventually higher speeds. Suppliers must develop the next generation while reducing the price and power consumption of the current one.
Standing still is rarely an option.
The Chips Inside the Modules
Ligent’s optical-chip business illustrates both the promise and difficulty of vertical integration.
Selling its own chips to external customers still generates only a small share of revenue. Optical chips contributed RMB28.9 million in 2025—just 0.3% of total sales—and the division recorded negative gross margins in both 2024 and 2025. The company said demand for older products weakened while new chips remained in development or production ramp-up, leaving revenue too low to cover fixed manufacturing costs.
The picture improved in the first half of 2026. External chip revenue rose to RMB85 million, and the segment reported a 40.7% gross margin after Ligent increased production of a higher-power continuous-wave laser chip introduced the previous year.
The turnaround is encouraging, but the chip operation should not yet be mistaken for a large independent semiconductor business. Its more immediate value may lie inside Ligent’s transceivers: internal chip capabilities can help the company coordinate the development of lasers, packaging and complete modules.
This is particularly important as the industry experiments with technologies such as silicon photonics, linear-drive pluggable optics and co-packaged optics. Each represents a different attempt to transmit more data while reducing the power and complexity required to move it.
The winner will not necessarily be the company with a single superior component. It may be the supplier that can combine chip design, packaging, manufacturing and customer qualification quickly enough to keep pace with data-center architecture.
A Global Company Assembled Across Borders
Ligent’s corporate geography is unusual for a Chinese technology manufacturer.
It operates production facilities in Qingdao and Jiangmen in China, Nonthaburi in Thailand and New Jersey in the United States. Its Chinese plants provide scale across transceivers, chips and network terminals. Thailand offers an additional manufacturing base for overseas deliveries, while the New Jersey operation gives the group production and technical capabilities closer to North American customers.
The company established a Silicon Valley research center in 2017, began building its Thai manufacturing presence in 2019 and opened a Singapore R&D center in 2026.
That footprint was assembled over many years, but it has acquired new importance as technology supply chains become more regional. Cloud companies increasingly want suppliers that can manufacture in more than one country, respond locally to product-development requests and reduce their exposure to tariffs or disruptions at a single production site.
The shift is already visible in Ligent’s sales. North American revenue reached RMB1.21 billion in the first half of 2026, almost twice the figure from the same period in 2025. China remained the largest market, but its share of group revenue declined as overseas sales grew faster.
Global production does not eliminate geopolitical risk. It makes the company more adaptable to it. Ligent’s next challenge will be converting that footprint into durable customer relationships without allowing the cost of parallel facilities to erode its margins.
The Concentration Behind the Growth
Fast-growing infrastructure suppliers often depend on a small number of very large buyers, and Ligent is no exception.
Its five largest customers contributed 55.8% of revenue in 2023. That proportion increased to 70.2% in 2025 and 71.8% in the first half of 2026. Its largest customer alone accounted for 21% of revenue during the latest period.
The prospectus does not name these customers, describing them instead as major cloud-service providers and telecom or network-equipment companies in China and overseas.
Concentration can be a sign of successful entry into the supply chains that matter. Large cloud operators buy at a scale few other customers can match, and their qualification processes can make supplier relationships relatively stable. But the same structure gives buyers considerable leverage. A delayed product approval, lost platform design or change in procurement strategy could have a disproportionate effect on revenue.
The industry’s pace compounds the risk. Optical products generally become cheaper during their commercial lives. A supplier must therefore increase volume, improve manufacturing efficiency or introduce higher-value products simply to defend its economics.
Ligent’s recent growth shows what happens when the product mix moves in the right direction. Its history in telecom equipment shows what can happen when a market matures and customer spending slows.
What Hong Kong Is Being Asked to Price
Ligent expects to receive approximately HK$5.45 billion in net proceeds from the IPO, before any exercise of the over-allotment option.
It plans to devote 52.9% of that amount to research and development, including the recruitment of engineers and the purchase of specialized equipment. Another 25.1% is earmarked for expanding transceiver and chip capacity and increasing factory automation. The remainder will support overseas expansion, acquisitions and working capital.
Those allocations reveal the scale of the wager. Ligent is raising money not simply to produce more of its current devices, but to keep moving through several generations of optical technology while building capacity in advance of demand.
The IPO is therefore a test of two propositions.
The first is that spending on AI infrastructure will continue expanding beyond processors into networking, optical components and other less visible layers of the data center.
The second is that Ligent can convert its unusual combination of Chinese manufacturing, North American optical-chip expertise and long-established telecom relationships into a lasting position in that market.
According to research commissioned for the prospectus, Ligent ranked fifth among specialized optical-transceiver manufacturers worldwide in 2025, with a 4% global revenue share. In China, it ranked third with a 10.1% share. These figures place it among the established challengers, but still behind a small group of larger competitors.
That may be the most useful way to understand the company approaching the Hong Kong market. Ligent is not an obscure manufacturer suddenly dressed in AI language. Nor is it yet the company setting the direction of the global optical industry.
It is a two-decade-old broadband supplier attempting one of the most consequential transitions in its history.
The AI race is usually described as a contest to build the most powerful chips. In practice, those chips must operate as a coordinated system—and that system depends on the speed of the light moving between them.
Ligent has spent years building the components that carry that light. Its IPO will help determine whether it can move as quickly as the market it now serves.
