AI computing power has driven explosive growth in the optical module market, with 800G and 1.6T technologies leading the industry transformation. Chinese companies occupy a dominant position in global competition. Coupled with the explosive growth in AI inference demand and the expansion of emerging application scenarios, the high prosperity of the optical module industry will continue in 2026.

I. Market Demand: 800G Volume Surge, 1.6T Deployment, Demand Forecast Revised Upward Beyond Expectations

A. 800G Optical Modules: Rising Demand and Breakthroughs in Technological Approaches

Shipments of 800G optical modules are projected to exceed 40 million units in 2026, with demand dominated by North America and by China. Meta, Google, Microsoft, and Amazon are the core purchasers, with Meta's demand reaching 10-12 million units, Google and Microsoft combined at approximately 12 million units, and Amazon around 5.5 million units. Domestically, ByteDance, Alibaba, and Tencent are gradually increasing their demand, with the three companies combined reaching several million units. Market confidence in 800G is strong, with Goldman Sachs raising its 2026 sales forecast from 25 million units to 33.5 million units, an increase of 58%. Technically, ByteDance is skipping traditional solutions and directly pursuing LPO technology; if successful, this could become a dark horse in the 800G field.

B. 1.6T Optical Modules: Ushering in the Year of Commercialization, Demand Doubles

2026 is projected to be the first year of commercialization for 1.6T optical modules, with global demand expected to reach 8.6 million to 20 million units. Nvidia is the key driver, with demand expected to exceed 5 million units in 2026. Google and Meta are expected to require approximately 4 million and 1 million units respectively. Driven by the AI computing power race , overseas tech giants have increased their 2026 procurement plans from 10 million to 20 million units. A single GB200 server requires 162 1.6T optical modules, further amplifying demand. Technologically, silicon photonics has become mainstream, holding a 72% market share. InnoLight Technology's self-developed silicon photonics chip boasts a 95% yield rate and reduces costs by 30% compared to traditional solutions. Tianfu Communication's 1.6T silicon photonics engine has a 90% yield rate, far exceeding the industry average of 75%.

II. Technological Trends: Three Leading Technologies Drive Accelerated Iteration

1). Silicon Photonics Technology: the mainstream technology for high-speed optical modules, with a steadily increasing penetration rate.

Silicon photonics technology, with its advantages of low power consumption and low cost, has become a core direction for 800G and 1.6T high-speed optical modules, and its penetration rate will continue to rise. Huawei released the world's first single-wavelength 1.6Tbps silicon photonics module , which adopts third-generation silicon photonics integration technology, reducing volume by 70% and power consumption by 40%. InnoLight Technology holds the world's number one market share in 800G silicon photonics modules , approximately 50%, and its 1.6T silicon photonics modules have already achieved mass production.

2). LPO Technology: Optimizing power consumption performance, becoming an important growth point in the market.

LPO technology achieves dual optimization of power consumption and performance by eliminating the DSP chip, perfectly meeting the needs of AI data centers. New Eoptolink's 800G LPO module reduces power consumption by 50% and has already secured orders from leading customers such as Meta and Amazon. It is projected that LPO module revenue will account for 15% of total revenue by 2026, becoming a significant growth driver in the optical module market.

3). CPO Technology: Cutting-edge exploration is underway, but large-scale application is still some time away.

CPO technology integrates the optical engine and switching chip into a single package, reducing data center power consumption by over 40%. TSMC's 3nm process micro -ring optical modulator technology achieves efficient conversion between optical and electrical signals, further reducing power consumption by 40%. However, CPO currently faces technical challenges such as alignment accuracy and thermal management , and only Broadcom can supply CPO switches, resulting in limited supply capacity. Large-scale application is expected to be available after 2027. Huawei HiSilicon, in collaboration with Flight Control Technology, launched the world's first 2.5D packaged CPO production line, with mass production planned for the fourth quarter of 2025, breaking TSMC's technological monopoly.

III. Competitive Landscape: Chinese Enterprises Dominate Globally, Leading Position Continues to Strengthen

Chinese companies have become a core force in the global optical module market, holding over 70% market share. Companies like InnoLight Technology , Hygon , Eoptolink, and Accelink Technologies have joined the ranks of the world's top tier. In LightCounting's 2023 global top 10 optical module rankings, Chinese manufacturers occupied seven spots, with InnoLight Technology topping the list for the first time . Leading companies secure demand through long-term contracts with key clients and continue to expand their market share based on their technological advantages. Industry resources are concentrating on leading companies with technological R&D and large-scale mass production capabilities, making the Matthew effect increasingly pronounced.

IV. Future Outlook: Multiple Positive Factors Support Continued High Industry Momentum

(1). AI inference demand becomes the core, and the increased allocation ratio strengthens growth resilience.

AI inference demand has surpassed training demand, becoming the industry's dominant growth driver. The industry has formed a positive cycle of "reduced technology iteration costs – application promotion – explosive growth in inference demand – data growth driving model iteration." Data from Huaxi Securities shows that overseas AI companies are experiencing rapid growth in demand, and token issuance is accelerating, driving high investment in AI capital expenditure. Meanwhile, the ratio of AI chips to optical modules continues to increase, with H100 reaching 1:3 and B300 reaching 1:4.5. Certain ASIC training clusters even reach 1:8. Even if AI chip sales slow down in the future, demand for optical modules will still show resilience.

(2). Expanding into new application scenarios opens up new market growth potential.

AI video generation and robotic computing power are becoming new growth drivers for the demand for optical modules. Social media platforms such as TikTok and YouTube are accelerating their AI integration. In 2025, AI content accounted for less than 2%, but it is expected to exceed 30% in 2026. This exponential demand for computing power directly drives the procurement of optical modules. Tesla's Optimus humanoid robot will be mass-produced in 2026 and needs to process multimodal sensor data in real time. Musk's "computing power matrix" plan requires computing power from billions of devices , reaching a scale of 50 million H100 equivalent computing power, further amplifying the market demand for optical modules.

(3). Dual support from policy and capital to help the industry achieve high-quality development

At the policy level, China's "East Data West Computing" project promotes the integrated layout of big data centers, releasing policy dividends for the optical module industry and driving the continuous growth of domestic demand. At the capital level, the number of financing events in the optical module field rebounded and the amount of financing increased significantly in 2024. Capital investment strategies have shifted from casting a wide net to focusing on leading companies with advanced technology and strong commercialization capabilities. Resources are concentrated to provide financial support for companies' R&D and capacity expansion, promoting the industry towards high-quality development.

Overall, the optical module industry in 2026 is in a period of dual benefits from technological iteration and explosive demand. 800G and 1.6T are driving market growth, core technologies continue to break through, Chinese companies have gained a competitive advantage in the global market, and the continued demand from AI inference and emerging application scenarios will ensure that the industry’s high prosperity will continue for a long time. Companies with core technologies, leading customer resources and mass production capabilities will continue to lead the market.