Several institutions now project that the global indium phosphide supply-demand gap will exceed 70% by 2026, while domestic manufacturers account for only about 10% of worldwide shipments, intensifying supply chain pressures that ripple through upstream and downstream segments. In an effort to identify the true bottlenecks at each stage, reporters conducted interviews with multiple listed companies spanning the indium phosphide value chain, including Yunnan Tin Co.,Ltd. (000960.SZ), Xingfu Electronics (688545.SH), Yunnan Lincang Xinyuan Germanium Industry Co.,Ltd. (002428.SZ), Yuanjie Semiconductor Technology Co.,Ltd. (688498.SH), Henan Shijia Photons Technology Co.,Ltd. (688313.SH), Suzhou Everbright Photonics Co.,Ltd. (688048.SH), and Sanan Optoelectronics Co.,Ltd. (600703.SH).
The findings reveal a domestic indium phosphide industry characterized by abundant resources but comparatively weak manufacturing capabilities. While upstream refined indium supplies are plentiful, capacity for high-purity indium remains insufficient. In the midstream, substrate and epitaxial wafer production are still ramping up, and high-speed optical chip makers continue to rely heavily on imported substrates. Overall, domestic production capacity remains limited relative to global demand. However, since the second half of 2025, the explosive growth of AI computing power has spurred coordinated efforts across the supply chain, accelerating the pace of domestic substitution.
Rigid Supply Gap Driven by the AI Computing Power Boom
As artificial intelligence computing demands rise exponentially, indium phosphide substrates are quickly becoming a focal point in global optical communications infrastructure. In July 2026, Michael Hurlston, CEO of U.S. optical chip leader Lumentum, stated publicly at the European AI Summit in Paris that the current supply-demand gap for indium phosphide substrates has already significantly surpassed that of mainstream memory devices such as DRAM and NAND flash. This signal not only highlights extreme tightness on the materials front, but also underscores the increasingly critical foundational role of indium phosphide in data center optical interconnect systems within the AI era.
An executive from a domestic indium phosphide substrate company further explained that the acceleration of the AI computing market has created a pressing challenge for high-speed optical module laser manufacturers and photodetector producers: achieving rapid data transmission to fully unleash the overall efficiency of graphics processing units (GPUs). As a result, the industry's mainstream solution now involves replacing traditional electrical signals with optical signals. Core components such as lasers and detectors, which form the basis of optical chips, require indium phosphide substrate materials. "The turning point came in the second half of 2025, when overall market demand for indium phosphide substrates began to climb," the executive added.
On the application side, representatives from high-speed optical chip companies such as Yuanjie Semiconductor Technology Co.,Ltd., Henan Shijia Photons Technology Co.,Ltd., and Suzhou Everbright Photonics Co.,Ltd. confirmed that AI data centers are currently the primary market driving demand for indium phosphide-based optical chip capacity. Regarding the global supply-demand landscape for indium phosphide substrates, the aforementioned executive noted that three foreign companies—Japan's Sumitomo Electric, Japan's JX Metals, and U.S.-based AXT—collectively hold more than 90% of the finished substrate market. All domestic manufacturers combined account for only about 10% of global supply. "There is considerable room for domestic substitution, but China's overall share of global production capacity remains relatively small."
Among domestic players, one of the largest is Xinyao Semiconductor, a subsidiary of Yunnan Lincang Xinyuan Germanium Industry Co.,Ltd., which currently operates at an annual capacity of roughly 150,000 wafers. "There's also Beijing Tongmei, a wholly-owned subsidiary of U.S.-based AXT, which has a relatively large production scale. Since it's not publicly listed, its exact capacity figures haven't been disclosed, and its products primarily supply its parent company AXT as well as export markets." The executive added that "from an industry-wide perspective, China's indium phosphide sector is still in its infancy. Historically, market demand wasn't this large. Now, computing infrastructure buildout is rapidly driving up demand for indium phosphide materials, but matching production capacity will take time."
Regarding the supply gap, research from China Galaxy Securities indicates that global demand for indium phosphide devices in 2025 stands at approximately 2 million wafers per year, while actual production capacity is only around 600,000 wafers annually. Third-party consulting firm Yole projects that global demand will rise to between 2.6 million and 3 million wafers per year by 2026, while effective capacity will only increase to roughly 750,000 wafers—leaving a supply-demand deficit exceeding 70%.
However, industry participants remain cautiously optimistic about the "70% supply gap" figure. A staff member from the securities department of Yunnan Lincang Xinyuan Germanium Industry Co.,Ltd. stated that the company does not base major operational decisions solely on market hype or a few research reports and third-party assessments. "Our expansion decisions hinge on two factors: first, the order volumes our downstream customers are likely to commit to; second, our own production and operational capabilities to execute such projects." The company's recent expansion plan adds an equivalent annual capacity of 300,000 wafers, driven by both the high probability of downstream demand materializing and an internal assessment confirming the company's ability to complete the project.
Behind the heated expectations, it's essential to distinguish between genuine demand and speculative narratives. AI computing center construction currently serves as the "fundamental driver" of indium phosphide substrate demand growth, and this trend shows signs of sustainability. According to Ministry of Industry and Information Technology data for January-February 2026, national investment in computing infrastructure grew 42% year-on-year. The "East Data West Computing" initiative's eight national hub nodes, provincial-level intelligent computing centers, and AI cluster projects by leading internet companies are all accelerating—these represent tangible, verifiable demand increments. By contrast, areas such as 6G and automotive LiDAR primarily reinforce long-term expectations. These applications remain at an early industrialization stage, with considerable distance before large-scale commercial deployment, and the market may be overinterpreting the near-term impact of these future applications on current supply-demand dynamics.
Substrate Breakthrough: Domestic Substitution Still in Its Climbing Phase
On the evening of August 21, A-share optical module leader Zhongji Innolight released its 2026 semi-annual report, showing revenue growth of 182.49% year-on-year and net profit growth of 241.7%. The company attributed the surge to increased capital expenditures by key customers investing in computing infrastructure. During the reporting period, Zhongji Innolight's prepayment balance jumped from RMB 134 million at the end of 2025 to RMB 954 million—a more than sixfold increase—primarily used to lock in upstream optical chip and substrate capacity in advance. A marketing executive from a domestic metal materials company publicly stated that lead times for indium phosphide substrates from downstream customers in 2026 generally range from 24 to 40 weeks, with some specifications requiring down payments of 30-50% just to secure a place in line.
According to the latest reporting, domestic high-speed optical chip shipments are accelerating noticeably, yet indium phosphide substrates—a critical material—still predominantly rely on overseas suppliers. Suzhou Everbright Photonics Co.,Ltd., which focuses on EML chips, does not have upstream indium phosphide substrate operations and instead purchases substrates externally to produce high-speed optical chips, positioning itself as a midstream optical chip manufacturer. The company's securities department stated that in response to the current global substrate shortage, it is validating and introducing domestic suppliers such as Yunnan Lincang Xinyuan Germanium Industry Co.,Ltd. and Beijing Tongmei.
Regarding substrate sourcing, a representative from Yuanjie Semiconductor Technology Co.,Ltd.'s board office said the company procures substrates from both domestic and overseas suppliers, though the supplier list is confidential. "The company adopts a diversified, multi-supplier strategy in substrate procurement to ensure stability of upstream raw material supply." Meanwhile, a representative from Henan Shijia Photons Technology Co.,Ltd. noted that all indium phosphide substrates required by the company are purchased externally; its CW-type active device products utilize indium phosphide substrate materials.
Sanan Optoelectronics Co.,Ltd. operates in indium phosphide epitaxial wafers and optical chips, achieving batch sales of related products. "All upstream indium phosphide substrates are sourced through market procurement; we don't produce them ourselves," a representative from the company's board office explained. The company currently has indium phosphide-based optical chip capacity of 2,750 wafers per month and epitaxial capacity of 6,000 wafers per month, with production lines continuing to expand.
Xinyao Semiconductor, under Yunnan Lincang Xinyuan Germanium Industry Co.,Ltd., is one of the few domestic companies capable of mass-producing indium phosphide substrates for optical communications. Regarding capacity plans, the company's board office stated that its current mass production focus is on 2-4 inch substrates, with existing capacity of 150,000 wafers per year. In April 2026, the company announced a RMB 189 million expansion project, planning to add an equivalent annual capacity of 300,000 wafers, with an 18-month construction period that includes a pilot line for 6-inch substrates at 6,000 wafers per year. "We've already supplied small quantities of 6-inch substrates to downstream customers and laboratories for trial and R&D purposes, but large-scale mass production will take more time."
As interest in the indium phosphide sector heats up, crossover players are entering the fray. In June, Xingye Technology, a leather business, announced plans to acquire Qingdao Li'ang for RMB 55 million to enter the indium phosphide substrate field. Xingye Technology stated that Qingdao Li'ang's products primarily focus on 2-inch and 3-inch substrates, while its 4-inch products remain in the validation stage—a high-barrier process that carries the risk of customer qualification failure. On production capacity, "Qingdao Li'ang currently has about 20,000 wafers per year, with expansion plans targeting 100,000 wafers annually. Trial production is expected to begin between Q4 of this year and Q1 2027, followed by gradual ramp-up," a source close to the target company said.
Another crossover entrant is Bojie Technology, which entered the upstream substrate arena by taking a stake in Zhuhai Dingtai Xinyuan. Dingtai Xinyuan can supply 2-4 inch substrates in batch quantities, while 6-inch products are still in R&D and sampling stages; specific capacity release figures have not been disclosed. As of Q1 2026, Dingtai Xinyuan had not yet achieved profitability. "At this stage, every company needs continuous optimization and improvement. It's not like building a factory in year one and producing qualified products in year two. From production line construction to downstream validation and batch supply typically takes 2-3 years or even longer," the substrate executive said, noting that each size upgrade—from 3-inch to 4-inch, and 4-inch to 6-inch—brings technical bottlenecks in processing R&D, including equipment modifications and proprietary temperature field control technologies. Every size transition requires recalibrating the entire process system and climbing the yield curve anew.
High-Purity Indium Refining: Low Volumes, Extremely High Barriers
From a supply chain perspective, the indium phosphide substrate manufacturing process involves synthesizing indium phosphide polycrystals from high-purity indium and phosphorus, growing single-crystal ingots, and then processing them into substrate wafers through cutting, grinding, and polishing. Domestic primary indium materials mainly come from by-products of lead-zinc-tin smelting, with domestic primary indium production accounting for over 70% of global output. However, high-purity indium and electronic-grade red phosphorus—core raw materials for indium phosphide single-crystal manufacturing—remain heavily import-dependent.
Reporting indicates that listed companies such as Yunnan Tin Co.,Ltd., Huaxi Nonferrous Metals, and Zinc Industry Co., Ltd. can consistently supply large volumes of 4-5N industrial-grade refined indium, with nearly 60% of sales going to the ITO target market, indicating ample resource reserves. Yet, with rising AI computing demand, supply of 7N and above semiconductor-grade ultra-high-purity indium is tightening. This class of high-purity indium is the core raw material for indium phosphide single-crystal growth, and domestic effective mass production capacity remains insufficient in the near term.
A representative from Yunnan Tin Co.,Ltd.'s board office stated that optical communication-grade indium phosphide substrates require extremely high indium purity—reaching 6N (99.9999%) to 7N (99.99999%) levels. The company's associated enterprise, Yunnan Tin New Materials, is advancing a high-purity indium project with a certain scale of 7N-level capacity and is currently expanding. Regarding 7N-level high-purity indium progress, on August 19, Yuguang Gold & Lead, a high-purity indium refining company, stated that it has built a pilot production line for 7N-grade material, currently in small-scale trial production with an annual output of 300 kilograms. Samples have been delivered to downstream customers for trial validation. Feedback is expected by end of 2026 or early 2027, after which expansion and project construction will be evaluated based on market conditions.
Another leading domestic indium refiner, Huaxi Nonferrous Metals, said its securities department began R&D work on 7N and above high-purity indium for semiconductor applications as early as 2025. To date, related products remain in the R&D and validation phase. "Although domestic refined indium resources are abundant, very few domestic companies can consistently produce 6N and 7N semiconductor-grade high-purity indium," said Huang Li (pseudonym), a senior analyst in Shanghai's minor metals sector. "Ordinary refined indium and semiconductor-grade high-purity indium may seem to differ by only a few nines of purity, but behind them lie completely different process systems. In recent years, with the surge in optical chip demand, domestic purification technology has advanced rapidly. Based on current project momentum, I believe domestic high-purity indium raw materials can achieve supply-demand balance within two years."
Huang further explained that another reason for the lag in domestic high-purity indium purification is the relatively small quantity of indium required for indium phosphide substrates, which directly reduces domestic companies' willingness to invest in this area. Currently, demand for 7N and above high-purity indium accounts for only about 10% of total indium material consumption. "The market keeps hyping the indium phosphide concept, and the public generally assumes metal consumption is massive, but actual material consumption is quite small." According to Huang, a single 2-inch indium phosphide substrate requires only about 1-2 grams of high-purity indium. Taking Yunnan Lincang Xinyuan Germanium Industry Co.,Ltd. as an example—one of the largest domestic substrate producers—its existing annual capacity of 150,000 wafers translates to only 300-400 kilograms of high-purity indium. This reality diminishes the incentive for upstream indium refiners to establish purification laboratories and equipment, entering a niche field with extremely high technical barriers. "Moreover, from a pricing perspective, ordinary refined indium currently trades around RMB 5,000 per kilogram, while 6N and above high-purity indium might only fetch around RMB 6,000. A premium of roughly RMB 1,000 isn't as attractive as one might expect," Huang added.
Comments