With the rapid expansion of AI computing clusters and large data centers, the optical communication industry is accelerating its iteration from 400G to 800G large-scale commercial use and 1.6T small-batch trial production. The core of upgrading high-speed optical modules lies not in simply increasing the speed, but in the technological innovation of core optoelectronic materials. In 800G/1.6T transmission scenarios, Faraday swirl plates, EML chips, silicon photonics chips, and DSP chips are the four core materials that affect module stability, transmission performance, integration, and cost, playing a crucial role in stable optical path transmission, electro-optical conversion, and integrated upgrades.

I. Core Components: The Cornerstone of High-Speed Optical Module Performance

1. Faraday Rotator: The “One-Way Security Gate” of the Optical Link.

As the core of the optical isolator, the Faraday rotator relies on the magneto-optical effect to achieve unidirectional transmission of optical signals to protect the laser. The 800G/1.6T module requires multiple compatible plates, and their performance directly determines the reliability of the module. Its technical barriers are concentrated in magneto-optical crystals (TSAG/TGG for high-end applications), micron-level processing, and high-precision coating. The customer certification cycle is long, and the industry has a strong monopoly.

2. EML Chip: The "Transmission Engine" of Traditional High-Speed Optical Modules

EML (Electro-Absorbed Modulated Laser) is the core transmitter chip of medium- and long-distance high-speed optical modules. It integrates DFB laser and modulator, is suitable for transmission over 2km and is widely used in telecommunications backbone networks and cross-regional data communication. Its core barriers lie in InP epitaxial growth, multi-quantum well design, etc. Only a few companies such as Lumentum can stably mass-produce it, and the technical threshold is extremely high.

3. Silicon Photonic Chip: The “Disruptive Solution” for the Integration.

Silicon photonics chips integrate optical components based on CMOS technology, which significantly reduces module power consumption, size and cost. They are suitable for 1.6T and CPO scenarios and are the first choice for short-distance interconnection within 500 meters in AI data centers. The core barriers are silicon photonics integration process and coupling packaging, while optical loss and coupling efficiency are the main technical challenges.

4. DSP Chip: The "Signal Center" of High-Speed Optical Modules

DSP (Digital Signal Processing) chips are the core signal processing brain of high-speed optical modules. They are responsible for encoding, decoding, equalizing, noise reduction, and distortion compensation of optical signals. They can effectively offset signal loss and crosstalk in high-speed, long-distance transmission and ensure stable data transmission. The core technological barriers lie in high-speed ADC/DAC, high-speed circuit design, algorithm architecture, advanced process tape-out, and optoelectronic co-adaptation capabilities.

II. Route Comparison: Each Has Its Own Focus; They Are Complementary and Symbiotic.

III. Supply Chain Pain Points: The Double Threat of Overseas Monopoly and Capacity Lock-in

1. Faraday Rotor: Overseas Monopoly + Supply Contraction, Huge Domestic Shortage

The global high-end Faraday Rotor market is monopolized by Japan's Granopt and the United States' Coherent. The two giants will reduce supply in 2025-2026, resulting in a global shortage of about 50% of high-end products. Meanwhile, domestic spiral blades are mainly low- and mid-range products, and there is no mass production of high-end TSAG spiral blades, and customer certification is lagging behind. It will be difficult to alleviate this gap in the short term.

2. EML Chips: Capacity Lock-in + Yield Gaps Make the Shortage of a Single Chip a Norm

In 2026, the global shortage of EML chips will exceed 150 million units. Overseas giants have locked up production capacity through long-term contracts, resulting in China's reliance on imports for most of the chips. China can only stably mass-produce 100G EML chips, while 200G chips are still in the verification stage and have a long expansion cycle. This shortage will continue until 2027.

3. Silicon Photonic Chips: Technical Barriers + Skyrocketing Costs, Bottlenecks in Large-Scale Mass Production.

The supply chain is vulnerable to imports and shortages have led to price increases. There is a shortage of foundry capacity and insufficient process yield. High-end chips are monopolized by overseas companies. Packaging and testing technologies are difficult and costly. The industry has a long verification cycle and high entry barriers. Coupled with the surge in AI demand and the sluggish expansion of production capacity, as well as external policy risks, the supply chain is easily subject to foreign control, making it difficult to promote domestic substitution.

4. DSP Chips: Consolidated Market Structure + Significant Technological Gap Makes Domestic Substitution Difficult

Overseas leading companies have long monopolized the global high-end DSP chip market. In 2026, industry demand surged while production capacity was locked up overseas, leading to continued supply shortages. Domestic production rates for this product category are low, with technological, process, and toolchain gaps. Coupled with challenges such as cost, certification, and production capacity, the progress of domestic substitution is slow. Furthermore, the continued surge in demand from fields such as AI and high-speed optical communication further exacerbates the supply-demand imbalance.

IV. Conclusion

The iterative upgrade from 800G to 1.6T is a critical window of opportunity for China's optical communication industry to break through the overseas monopoly and achieve independent breakthrough. Currently, the industry has formed a pattern of domestic substitution with silicon photonics leading the way and DSP chips, Faraday Rotor and EML chips catching up at an accelerated pace.

In the next 1-2 years, by tackling key technologies for four core materials, optimizing the layout of high-end production capacity, building an industrial collaborative ecosystem, and strengthening policy and talent support, the domestic optical module industry chain will break free from the "bottleneck" dilemma and achieve a leapfrog development from technology follower to global leader as technological breakthroughs and production capacity are realized.

Finally, ETU-Link is constantly iterating and breaking through, accumulating strength in new areas, and we look forward to in-depth cooperation and resource sharing with industry leaders in new fields. Feel free to discuss business and collaboration anytime!