Economic Observer reporter Zheng Chenye

As one of the most controversial concepts in the A-share market since 2026, the stock price of 'Yi Zhongtian' is currently undergoing a period of significant adjustment.

"Yi Zhongtian" refers to Xin Yisheng (300502. SZ), Zhongji Xuchuang (300308. SZ /03308. HK), and Tianfu Communication (300394. SZ). The first two companies produce optical modules (devices that convert electrical signals to optical signals in data centers), while Tianfu Communication supplies optical components to these optical module manufacturers.

On November 6th, the three stocks associated with "Yi Zhongtian" all experienced a decline of over 12% during trading, with a cumulative decline of up to 65% compared to their peak prices in June. On November 3rd earlier in the month, the decline for these same three stocks also exceeded 9%.

The reasons for this round of stock price adjustments are quite complex, including excessive gains in the earlier period, the technological competition brought about by NVIDIA's advancement of CPO (which involves encapsulated optics and may potentially replace some optical modules), and the expectation of a slowdown in the capital expenditure growth of cloud service providers.

On November 2nd, American chip company Micron Technology ( Marvell ) announced the acquisition of American optical device manufacturer Lumentum for approximately $115 billion. Both companies are major suppliers to optical module manufacturers, which raised concerns in the market about increased supply concentration upstream. Subsequently, there were also reports of "overseas customers canceling orders" in the market.

On the evening of November 3rd, Wang Jun, Vice President and Secretary of the Board of Directors at Zhongji Xuchuang, denied rumors of customers significantly reducing their orders during a teleconference with investors, stating that "the vast majority of customers have not lowered their demands or orders." Wang Jun also mentioned that the acquisition of Maiwei Technology "has no impact on our overall supply structure or the future competitive landscape."

However, looking at the third-quarter reports of optical module companies, there is a trend of slowing performance – New Yisheng’s revenue and net profit in the third quarter of 2026 decreased by 12.98% and 19.07% respectively compared to the previous quarter, and Tianfu Communication’s net profit also fell short of market expectations.

There are objective reasons for the slowdown in growth. The floods in Thailand in late September delayed the shipment from Xinyisheng's factory there, but the impact of material shortages seems to be even greater. Wang Cheng, the deputy general manager and secretary of the board of directors at Xinyisheng, stated at the company's performance meeting that raw materials had a certain impact on the company's production and deliveries in the third quarter of 2026.

The amount of materials that can be obtained determines how many orders can be fulfilled. The problem of material shortages is more troublesome for optical module manufacturers than a decline in stock prices.

The most critical component of an optical module is the chip – the optical chip is responsible for photoelectric conversion, while DSP (digital signal processing chips) and other electrical chips are in charge of handling high-speed electrical signals. According to information obtained by the reporter, both types of chips are currently in short supply, with the laser chips, which are responsible for emitting light, being particularly in demand.

There are mainly two approaches that optical module manufacturers take in response to shortages of upstream materials. First, they lock in materials from their suppliers. For example, as of the end of September 2026, the prepaid payments of Zhongji Xuchuang and NeoPhotonics increased by 1068.43% and 7695.01% respectively compared to the beginning of the year, with most of this amount being prepaid for materials from upstream suppliers. Second, they increasingly adopt silicon photonics solutions, using fewer laser chips. In traditional solutions, laser chips are responsible for both emitting light and loading data onto the optical signal, with one chip required for each transmission channel. However, with silicon photonics solutions, the function of data loading is separated and entrusted to dedicated silicon photonics chips, allowing a single laser chip to serve multiple channels, thereby reducing the consumption of laser chips.

So-called silicon photonics involves manufacturing optical devices on silicon-based materials using semiconductor processes, and silicon photonics chips are a type of optical chip. Such chips are produced by wafer foundries, which use specialized silicon photonics wafers for their fabrication.

As a result, with more optical module manufacturers adopting silicon photonics solutions, silicon photonics wafers have become a new bottleneck in the upstream of optical modules, following DSP and lasers, leading to a scarcity of foundry production capacity. Public information indicates that the production capacities of two major foundries, Israel's Gaotar Semiconductor and American GlobalFoundries, are already booked beyond 2027.

Switch to silicon photonics solutions

Optical module manufacturers are turning to silicon photonics solutions, as the traditional laser chips EML (electrically absorbed modulated lasers) that were previously used are no longer sufficient.

EML integrates light emission and modulation (loading data onto optical signals) on a single indium phosphide chip. In July 2026, Lumentum President and CEO Hirston publicly stated that even if Lumentum and its competitor Coherent combined their efforts, they still could not meet the laser demands of customers such as NVIDIA.

Silicon photonics solutions separate light emission from modulation. Silicon material itself is not efficient at emitting light, so indium phosphide laser chips are still used as the light source, but these are replaced with continuous-wave lasers ( CW ). These laser chips provide stable light that can be split for use by multiple channels after separation. For example, a 1.6T optical module has 8 channels; under the EML scheme, 8 EML lasers would be required, whereas with silicon photonics technology, 1 CW laser capable of handling 2 channels is sufficient, and only 4 lasers are needed in total.

Moreover, the amount of lasers used can be further reduced. Wang Jun stated during a teleconference on November 3 that if the proportion of silicon photonics solutions continues to increase and the modulation capabilities of silicon photonics chips are further enhanced, "it is possible to use 1 to drive 4 instead of 1 to drive 2, which would allow for further cost reduction." In other words, under such circumstances, the usage of CW lasers in 1.6T optical modules could be reduced by another half.

In addition to the reduction in materials used, the total cost of silicon photonics solutions is also lower. According to Xihe Technology, a silicon photonics chip design company, as mentioned in its prospectus, based on the calculations of optical module manufacturers, the total cost of silicon photonics solutions can be reduced by nearly 20% under the same rate and distance conditions.

In fact, Wang Xiaoli, Vice President and Chief Financial Officer of Zhongji Xuchuang, also cited the increased penetration rate of silicon photonics solutions as one of the reasons for the company's rising gross margin during a Q3 earnings conference call. In the third quarter of 2026, the company's gross margin was 46.6%, an increase of 0.2 percentage points from the previous quarter.

In addition, silicon photonics solutions are also more power-efficient. Xihe Technology stated in its prospectus that these solutions have better power consumption and reliability compared to traditional ones. This is particularly relevant in the United States. A senior investor who has long been following the field of optical communications told reporters that the market generally views the Democratic Party's victory in the mid-term elections in the US as a negative for optical modules, but many people do not realize that silicon photonics solutions are actually more energy-efficient. Lin Meibing, the chief analyst at ChipSay ICTIME, also told reporters that in times of tight electricity supply, customers are more sensitive to the power consumption of optical modules when making choices, which is advantageous for silicon photonics solutions.

The proportion of silicon photonics solutions is also continuously increasing. As early as April 2026, Zhongji Xuchuang stated on its investor interaction platform that the proportion of silicon photonics solutions in its 800G and 1.6T products had already exceeded 50%. A communications analyst from a large securities firm in South China told reporters that in the entire industry's 1.6T optical modules, the proportion of silicon photonics solutions was around 60% to 70% in 2026, and it may reach around 80% in 2027.

As the proportion of silicon photonics solutions continues to increase, the pressure falls on the wafer foundry process.

Silicon photonic wafers are in short supply.

There are not many foundries that can produce silicon photonic wafers, and it will also take time for the additional production capacity to be put into use.

Wafers are the basic materials for manufacturing chips. Foundries create circuits on the wafers, and after cutting, bare chips are obtained. Ordinary chips have circuits designed to process electrical signals on their wafers, while silicon photonic wafers are used to produce devices such as modulators and detectors that process optical signals, and the manufacturing process is different from that of ordinary chips.

Lin Meibing told reporters that the manufacturing process of silicon photonics chips is complex, and the threshold for yield is high. There are not many foundries capable of mass production on a large scale. Currently, this capability is mainly concentrated in a few manufacturers, including Israel's wafer foundry GlobalWafers and America's GlobalFoundries.

On November 2, Elvanger, the CEO of Gaota Semiconductor, stated at the earnings conference that the company's silicon photonics chip foundry capacity contracts with customers have been extended until 2029. On November 4, Brian Krzanich, the CEO of TSMC, also mentioned at the earnings conference that the company's silicon photonics chip foundry capacity has actually been sold out, with the capacity for 2027 already fully booked, and the capacity for 2028 also being gradually reserved.

With limited production capacity, downstream customers can only secure their spots in advance. In May 2026, Gaota Semiconductor announced that it had signed contracts with its largest clients, resulting in $1.3 billion in revenue from silicon photonics wafer foundry services for 2027. The company also received a prepayment of $290 million from customers to reserve production capacity.

Contract manufacturers are also expanding production. Elvanger stated that by the third quarter of 2027, the wafer placement (putting wafers into production lines) capacity of Gaota Semiconductor's silicon photonics wafers will be tripled compared to the level of the third quarter of 2026, and most of this additional capacity already has customers. Brin also mentioned that GlobalFoundries plans to increase the production capacity of silicon photonics chips by about three times in 2027.

A later batch of new production capacity is not expected until after 2028. GaoTa Semiconductor aims to start producing at its newly built factory in Japan in 2029, while TSMC, a wafer foundry in Taiwan, China, will only begin installing new production capacity by the end of 2028.

After the new factory is completed, the silicon photonics chips cannot be immediately transferred there. A market manager from a optical module manufacturer in South China told an Economic Observer reporter that when silicon photonics chips are moved to a new factory, they have to go through the process of process verification again, which typically takes about a year. During the interview, the reporter also learned that previously, some silicon photonics products skipped this process in order to meet deadlines and were directly produced in the new factory, resulting in unsatisfactory yield rates for some batches and affecting downstream deliveries.

It is difficult to increase production capacity in the short term. Lin Meibing believes that how much production capacity can be obtained at present depends on whether contracts have been signed in advance.

Currently, many of the silicon photonic wafers used in mainstream pluggable optical modules are 8 inches in size. According to the plans of the foundries, the total production capacity of 8-inch wafers is basically sufficient to meet this demand. However, by 2027, most of the production capacity has already been reserved through agreements. In other words, the manufacturers who are short of production capacity are those who did not sign contracts in advance.

12-inch silicon photonic wafers are in even greater demand, as they are primarily used in next-generation products such as NPO (near-packaging optics). According to Lin Meibing, based on manufacturers' demand plans for 2027, the current production capacity is far from sufficient. A new round of capacity expansion is focused on 12-inch wafers, but the construction cycle is long, so supplies will still be tight until the end of 2027. However, Lin Meibing believes that this supply shortage may not persist in the long term, as there are already a dozen manufacturers engaged in silicon photonic chip manufacturing. Once new entrants start mass production, supplies will gradually increase.

In fact, newly entered contract manufacturers have already begun to compete for 12-inch orders. Industry insiders told reporters that there are already differences in quotes; some contract manufacturers are quoting around $10,000 per chip, while others are quoting around $8,000.

But before the newly added production capacity is actually released, whether one can claim that capacity will depend on the position of the relevant manufacturers among the contract manufacturers.

The gap between self-developed and externally purchased solutions

The position of relevant manufacturers among contract manufacturers is related to where the silicon photonics chips come from.

Silicon photonics chips come from two sources: one is that optical module manufacturers design them themselves and then commission foundries for production; the other is that optical module manufacturers purchase them from third-party chip companies. The aforementioned communication analyst believes that in the 800G era, optical module manufacturers competed on delivery and yield, but in the 1.6T era, silicon photonics solutions have become the mainstream, and chip design capabilities as well as foundry production capacity have begun to affect market share.

Headlight module manufacturers have already progressed to the stage of chip design. For example, in response to inquiries from the Shanghai Stock Exchange, Xihé Technology stated that light module manufacturers such as Zhongji Xuchuang and Xinyisheng are "expanding their capabilities in silicon photonics integrated chip design." These chips are used to complement their own light module products and are not sold externally. Public information also indicates that Pai Xi Technology, a subsidiary of Zhongji Xuchuang, and Alpine acquired by Xinyisheng are both engaged in the research and development of their own silicon photonics chips.

Optical module manufacturers with the capability to develop their own silicon photonics chips will place orders directly with foundries. During interviews, several industry insiders stated that leading optical module manufacturers hold a significant share of silicon photonics chip orders at foundries and are given priority in capacity allocation over third-party chip companies. If optical module manufacturers purchase chips from third-party chip companies, the cost will increase accordingly. For reference, the gross profit margin of the third-party chip company Xihe Technology in the first half of 2026 was 59.24%.

Of course, as the proportion of head optical module manufacturers developing their own silicon photonics chips increases, the pressure on third-party chip companies is also growing. In the first half of 2026, Xihe Technology's largest customer accounted for 55.55% of sales. The company noted that leading optical module manufacturers already possess the capability to develop their own silicon photonics integrated chips. If downstream manufacturers all begin large-scale self-development and change their strategy of purchasing externally, their customer demand and business performance may be adversely affected.

Xihe Technology is a relatively large third-party chip company that adopts a fabless model (it only focuses on chip design, leaving manufacturing to foundries). According to data from Frost & Sullivan, its global market share of silicon photonics integrated chips was approximately 13% in 2025. Public information indicates that Xihe Technology's orders are growing rapidly: in the first half of 2026, its revenue reached 569 million yuan, which has already surpassed the 461 million yuan for the entire year of 2025; as of the end of August 2026, it had 4.551 billion yuan in outstanding orders.

With the rapid growth of outstanding orders, the wafer demand of third-party silicon photonics chip companies is also increasing rapidly in 2027. During interviews, reporters learned that the wafer demand of some third-party silicon photonics chip companies has clearly exceeded the production capacity guaranteed by their contracts with foundries.

The market manager of the aforementioned optical module manufacturer also mentioned that their company mainly relies on purchasing silicon photonic chips from external suppliers. When the foundry capacity is tight, even chip manufacturers have to wait in line, and their company ends up waiting after the leading manufacturers, resulting in longer lead times for receiving the chips.

In fact, domestic silicon photonics wafer foundry platforms have already begun to emerge, although their cumulative scale is not yet large. According to data from Frost & Sullivan, as of the end of April 2026, YueXin Semiconductor (301660. SZ), a wafer foundry, is one of the few domestic companies with the capability for large-scale production of 12-inch silicon photonics wafers. Its products include 400G, 800G, and 1.6T silicon photonics modules, with a cumulative number of wafers produced exceeding 7,000.

In this regard, Lin Meibing stated that there is already a foundation in China for the design of silicon photonics chips. However, there are not many manufacturing facilities capable of mass-producing 12-inch chips. Some third-party chip companies have begun to outsource their production to domestic 12-inch platforms, but for domestic wafer foundries to undertake high-end product manufacturing, there is still a need for time to verify the manufacturing processes and improve yield rates.

(Author: Zheng Chenye)