Today, optical modules are rapidly iterating from 800G to 1.6T and 3.2T, and optical chip materials and integration routes have become the core points of competition in the industry.

Indium phosphide (InP), electro-absorption modulated lasers (EML), silicon-based optoelectronics (SiPho/silicon photonics), and thin-film lithium niobate (TFLN) are not parallel substitutes, but rather a progressive supporting chain of "light source substrate - integrated high-end device - cost-reducing hybrid solution - next-generation high-performance modulation carrier"!

I. Four Core Concepts

II. Complete Light Emission and Modulation Link: Logic of the InP→EML→Silicon Photonics

The core process of optical modules to complete data transmission consists of two steps: 1. Generating laser light in the communication band (light emission, which can only be achieved by InP); 2. Controlling the light intensity with electrical signals (modulation, which is divided into three technical paths). The relationship between the four steps is entirely based on the division of labor of "light emission-modulation".

Path 1: All InP Monolithic Integration – EML Solution (High-end Long-Distance Route)

The entire transmission link relies entirely on InP, with no heterogeneous bonding and no external light source.

DFB multiple quantum wells are epitaxially grown on InP substrates to directly generate 1550nm narrow linewidth continuous laser light upon power-up.

EAM electroabsorption modulation region is etched on the same wafer, and optical signals are quickly switched on and off by voltage.

The chip integrates waveguides, detectors, and optical amplifiers, resulting in extremely low optical path coupling loss and minimal crosstalk.

Applicable Scenarios: long-distance interconnection for cloud vendors, coherent optical modules, 1.6T DR/FR series; advantages are the best signal quality and the longest transmission distance; constraints are the scarcity of InP capacity and high cost, which prevent large-scale deployment.

Path 2: InP emission + silicon modulation – Silicon photonics hybrid solution (cost reduction route for short to medium distances)

By separating light emission and modulation, and replacing most InP devices with silicon, scarce substrates can be saved.

Independent InP CW lasers are manufactured separately, producing pure, continuous laser output;

The laser beam is coupled into the silicon photonic chip via optical fiber;

The silicon-based MZM/micro-ring modulator changes the optical phase to complete the signal loading; the receiver relies on a germanium detector on silicon to receive the light.

Applicable scenarios: The industry positioning is a cost-effective solution for short-pitch rack interconnect of 800G and below. In 2026, the market share of silicon photonics modules has exceeded 50%.

III. Core Advantages of Thin-Film Lithium Niobate (TFLN) and its Complementary Roles to Existing Solutions

TFLN does not replace InP, but rather optimizes the modulation stages of silicon photonics and EML, offering three overwhelming performance advantages over silicon and InP modulators:

(1) Ultra-high modulation bandwidth: the electro-optic coefficient reaches 31pm/V, the traveling wave modulation bandwidth exceeds 110GHz, and a single channel can support a rate of over 240Gbaud, far exceeding the 60-70GHz upper limit of silicon photonics, and is compatible with 3.2T and above ultra-high-speed modules.

(2) Low power consumption and low drive voltage: The half-wave voltage is reduced to less than 1V, which greatly alleviates the high-frequency heat generation problem, solves the power consumption bottleneck of silicon photonics, and meets the low energy consumption requirements of AI data centers;

(3) Excellent signal linearity: no carrier dispersion distortion in electro-optic response, low signal loss under high-order modulation, and better long-distance transmission performance than InP-EAM modulators.

(4) The shortcomings of TFLN are: its process maturity is lower than that of silicon photonics, and its heterogeneous bonding packaging suffers from coupling losses; it only has modulation capabilities and must be used with an InP light source, and cannot form an independent chain.

IV. Summary

(1) Underlying hierarchy: InP is the basic light source material for all technical routes; EML is an integrated high-end device of InP; silicon photonics and TFLN are both modulation carriers that "control light but do not emit light", and must be used in conjunction with InP;

(2) The division of labor logic is as follows: EML = all-InP high-end integrated; Silicon Photonics = InP light source + silicon modulation (cost reduction and short distance); TFLN = InP light source + lithium niobate modulation (ultra-high speed and low power consumption).

(3) The core principle remains unchanged: silicon and thin-film lithium niobate can only optimize the modulation process, but cannot overcome the physical limitations of light emission. The essential position of InP at the bottom of the optical communication industry chain cannot be replaced.

(4) Iteration direction: During the 1.6T transition period, multiple routes coexist, and in the 3.2T era, "InP + thin-film lithium niobate" will become the mainstream technology combination for high-end optical modules.

Looking at the evolution of the four core technologies, it is clear that whether it is the mainstream EML and silicon photonics solutions or the TFLN technology for the 3.2T era, they all rely on InP, which is the lifeline for the high-speed iteration of optical modules.