Market rumors are rife, the trillion-yuan market cap leader is questioned, and Zhongji Chuangxun speaks out: Optical modules are by no means simple assembly

Innolight (300308.SZ), whose share price has recently fluctuated, has provided a focused response to market concerns.

On July 5, Innolight successively addressed multiple market rumors on the investor interactive platform, including Corning's Glass Bridge technology, supply chain stability, and the claim that "optical modules are only midstream assembly", in an attempt to dispel market doubts.

Regarding Corning's newly released Glass Bridge technology, Innolight stated that this solution belongs to a new technology for the optical coupling components inside CPO (co-packaged optics), and is not a substitute for optical module products. Even if it becomes mainstream in the future, the company's existing diversified technology layout will be compatible with it.

In response to the rumor that "upstream glare sheets are under embargo", Innolight explicitly denied it, stating that current core material procurement channels are normal, suppliers continue to provide support, and there is no situation of restricted supply. The company also responded to the questioning that it is "just a midstream assembly plant", emphasizing that the system design, integration optimization, and manufacturing processes of optical modules are all core technologies, by no means simple splicing or assembly.

Innolight is the leading optical module manufacturer on the A-share market. Its share price has soared since May last year, and this year it has become the second stock trading above 1,000 yuan on the ChiNext board. In the first half of this year, Innolight's share price rose by more than 100% cumulatively. However, from June 23 to July 6, Innolight's share price fell from 1382.33 yuan per share to 1098.92 yuan per share, with a cumulative decline of over 20% in 10 trading days.

The focus of market discussion is not just the Glass Bridge itself, but whether the two-year sustained high prosperity of AI optical modules can continue. In addition, as CPO gradually matures, whether the value distribution of the optical module industry chain will change is also a market focus.

Regarding the above issues, a reporter from *Time Weekly* called Innolight's securities department on July 6. The respondent stated that an interview letter needed to be sent. Subsequently, the *Time Weekly* reporter sent an interview letter to Innolight's public email, and no reply was received as of press time.

Profits May Be Hard to Earn Relying Solely on Optical Module Assembly

In fact, Innolight has always maintained a relatively cautious attitude towards the impact brought by CPO.

In May this year, some investors raised the question of whether Innolight would still be a core enterprise in the industry if CPO largely replaced pluggable optical modules.

In response, Innolight did not give a clear answer, only stating that the company has already made technical layouts in fields such as NPO and CPO, and will assist key customers in customizing and developing related products. Closely following the iteration trends of the industry and customers, the company can meet the demands of data centers for higher bandwidth density and lower power consumption by increasing R&D investment and accelerating technology upgrades, and is confident in continuing to maintain and consolidate its advantages and position in the industry.

CPO is regarded as an important direction for next-generation data center optical interconnection, and has also become a key variable for the market to re-evaluate the long-term value of the optical module industry.

In its 2025 annual report, Innolight gave a very clear definition of CPO (co-packaged optics): co-packaging the switching ASIC chip and the silicon optical engine on the same high-speed motherboard to reduce signal attenuation, lower system power consumption, cut costs, and achieve high integration. The company also judged that CPO is currently going through the transition from technological breakthroughs to early commercialization, and it is expected to take some time for industry standards to be formed. However, the mature application of CPO may bring major changes to the ecosystem of the optical module industry chain.

Nevertheless, the market is more concerned about whether the value of the industry chain will be redistributed after CPO becomes widespread.

In recent years, AI optical modules have become one of the fastest-growing links in the industry chain. Optical module manufacturers have shared the dividends brought by AI infrastructure construction relying on product design, manufacturing processes, and large-scale delivery capabilities. However, with the further integration of optical engines and switching chips, industry insiders generally believe that the profit distribution of the industry chain may change in the future.

A research report from Zheshang Securities pointed out that as the penetration rate of CPO accelerates, the value is concentrated and the certainty is strongest in the upstream core chip and advanced packaging tracks. From the perspective of industry chain cost breakdown, the value of CPO complete machines is highly concentrated in the upstream core barrier links, among which the three sectors of silicon optical PIC integrated wafers, ELS external light sources, and 2.5D/3D advanced packaging together account for more than 75% of the cost share, making them the core value highlands of the industry chain.

Judging from the currently public information, the layouts disclosed by Innolight are mainly focused on the R&D of products such as optical modules and NPO, and it has not publicly introduced its business layouts in fields such as silicon optical PICs, ELS external light sources, or advanced packaging.

"In the past, profits were made by assembling optical modules, but in the future, most of the profits will gradually flow to silicon optical chips and glass packaging substrates," Zhang Xiaorong, dean of the Deep Tech Research Institute, pointed out to the *Time Weekly* reporter. In the future, if CPO becomes mainstream, manufacturers that only rely on assembling finished products will earn less and less; only by independently developing silicon optical chips and manufacturing the entire set of CPO components can they maintain high profits; the gross profit margin of those who only assemble externally purchased parts will continue to decline.

Zhang Xiaorong emphasized that enterprises that develop silicon optical chips independently will have the highest say, followed by enterprises that produce glass substrates and wafer packaging, then manufacturers that can manufacture complete sets of CPO optical components, and the profits of traditional optical fiber accessory manufacturers will continue to shrink.

Glass Bridge Has Limited Short-Term Impact

After Corning released the Glass Bridge technology, what the market is most worried about is not this new type of optical interconnection component, but whether the industrialization pace of CPO will be advanced as a result.

On June 24, Corning released the Glass Bridge optical interconnection component in Seoul, which is positioned to connect optical semiconductors and optical fibers, targeting the emerging co-packaged optics (CPO) and glass-core semiconductor packaging architectures.

"The greatest significance of (Glass Bridge) is that it brings the tricky optical coupling process of CPO from precision mechanical assembly into the era of semiconductor wafer manufacturing, which is expected to reduce coupling costs by 20%-30%, making it an enabling tool for CPO," Lu Hong, founder of M&A Master, told the *Time Weekly* reporter.

A research report from Galaxy Securities pointed out that Glass Bridge uses ion-exchange waveguides inside glass to achieve mode field transition, supports 30μm-level high-density channels with coupling loss <2dB and enables detachable passive alignment of TMT ferrules, which essentially pushes CPO from being optically transmissible in the laboratory to being operable and maintainable in computer rooms.

Zhang Xiaorong further pointed out that in the past, manufacturing CPO required manual fine alignment of optical fibers, which was difficult to do, produced many defective products, and could not be repaired when broken. The Glass Bridge is equivalent to replacing a cluster of optical fibers with a special glass sheet, directly manufacturing optical paths in batches as a whole, without the need for fine adjustment, making production easier, reducing costs, and enabling disassembly, replacement, and maintenance, solving the most difficult mass production problem of CPO.

Of course, the short-term impact of this technology is still limited. Lu Hong believes that "in terms of industrialization pace, (Glass Bridge) is expected to advance the CPO penetration rate by 1-2 years in the medium and long term, but the short-term impact is limited, and customer verification and production line ramp-up are still required."

In fact, industry insiders generally believe that at least in the next few years, CPO and pluggable optical modules will still coexist for a long time. CPO is mainly applied to ultra-short-distance high-speed interconnection inside AI servers and within cabinets, while scenarios such as cross-cabinet and cross-data center still have stable demand for pluggable optical modules, and the two technical routes are difficult to achieve full substitution in the short term.

Therefore, what Glass Bridge changes is more the pace of industrial evolution, rather than the short-term competitive landscape. For Innolight, the focus of market attention is not whether its existing business will be quickly replaced, but whether the company can continue to occupy an important position in the industry chain in the next round of technological upgrading.

Lu Hong believes that Innolight's valuation logic is switching from cyclical hardware (20-25 times PE) to an AI computing power platform, but the upper limit depends on chip control power and solution barriers. If it successfully binds chips upstream, its valuation is expected to rise to 28-35 times PE; if it remains mainly focused on packaging and integration, it may maintain 22-28 times PE, and will continue to bear the valuation discount brought by the game of technical routes.

On July 6, Innolight closed at 1098.92 yuan per share, down 1.53%, with a total market value of 1.23 trillion yuan.

This article is from the WeChat public account "Time Weekly" (ID: timeweekly), author: Zhu Chengceng, published with authorization from 36Kr.