Author: He Lvheng, Editor: Li Motian
In the current capital market, the hottest chemical elements, besides silicon and copper, are probably indium, or more precisely, a compound named indium phosphide (InP).
Indium phosphide sits at the very upstream of the optical transceiver supply chain. High-speed optical transceivers and co-packaged optics (CPO) require optical chips, and more than half of these optical chips are made of indium phosphide.
The AI data center construction boom has driven both volume and price increases for high-speed optical transceivers. The domino effect has reached the very upstream, pushing indium—a rare chemical element not even covered in the nine-year compulsory education curriculum—into the spotlight. From last year to this year, the secondary market has been driven by one core theme: "supply falls short of demand."
At an industry summit earlier this month, the CEO of optical chip giant Lumentum stated that the shortage of indium phosphide is even more severe than that of memory. "I don't think our capacity, along with Coherent's, will be able to meet the needs of Nvidia and other customers [1]." These two companies hold over an 80% share in high-end indium phosphide optical chips.
On the other hand, Nokia, one of Lumentum's major clients, has been pushed to step in and build capacity itself. Last year, it acquired Infinera to secure its California indium phosphide wafer fab, and recently made an additional acquisition of NXP's Arizona facility, aiming to convert it into an indium phosphide optical chip production line.
Complaints from downstream companies are emerging one after another, putting the pressure squarely on indium phosphide wafer fabs. Japan's Sumitomo and the U.S.'s AXT, which monopolize the supply, have become the targets of criticism.
However, AXT refuses to bear the pressure alone. In recent earnings calls, it has repeatedly emphasized that the real culprit behind the indium phosphide shortage is someone else.
Natural Endowment and Export Controls
In May, when Trump visited China, Coherent's CEO booked a seat on the accompanying Airbus even earlier than Jensen Huang.
Coherent is one of the world's largest indium phosphide optical chip suppliers, and Nvidia's CPO uses Coherent's indium phosphide lasers. To ensure supply, Nvidia not only made a strategic investment of $2 billion in Coherent, but when Coherent's new factory topped out in June, Jensen Huang personally attended the foundation-laying ceremony, encouraging his good partner to step up capacity expansion.
Coherent's trip to China this time was ostensibly to "promote Sino-U.S. optical chip cooperation," but the real purpose is well known both inside and outside the industry: they want the goods.
The production of indium phosphide wafers can be simply divided into three steps: smelting of primary indium, purification of indium, and manufacturing of indium phosphide wafers. Coincidentally, China has choked off all three links to varying degrees.
First is the smelting of primary indium.
China's biggest advantage lies in its vast territory and abundant resources. The global reserves of indium are only 50,000 tons, of which only half is minable. Among the proven global indium reserves, China accounts for over 70% [2].
Another advantage is its world-leading scale in metal smelting.
Since indium is dispersed in the earth's crust (no rich deposits have been found to date) and is difficult to mine independently, the vast majority is produced as a by-product during the smelting of zinc/tin. Its capacity actually depends on the supply and demand of zinc/tin, rather than following the price fluctuations of indium itself.
It is like the dregs left over from making tofu; no one would build a tofu factory specifically to get the dregs.
But China produces the most "tofu," so it naturally has more "tofu dregs" than anyone else. China's output of primary indium accounts for over 50% of the global total. At its peak, out of the 400 tons of indium produced globally each year, 80% was exported from Yunnan, China [3].
Next is the purification of indium.
Primary indium (purity 95%-99%) is purified to refined indium (4N-5N grade, purity 99.99%-99.999%), and then further purified to high-purity indium (6N grade, purity 99.9999%) before it can be used to manufacture indium phosphide optical chips.
China's output of refined indium accounts for 70%-80% of the global total, making it the absolute hegemon. High-purity indium of 6N grade and above is still dominated by Japanese manufacturers, with Dowa and JX Metals occupying most of the market. However, the primary indium used for their purification almost entirely comes from China.
Finally, there is the indium phosphide wafer, where the situation is rather special.
It is inherently China's weakest link, but fortunately, there is Beijing Tongmei, which voluntarily "naturalized" to China.
Beijing Tongmei is the operating entity of the U.S. company AXT in China, and it is currently the world's second-largest indium phosphide wafer supplier (first is Japan's Sumitomo). Back in the day, to be closer to the production sites of primary indium and save on freight costs, they stubbornly moved their entire indium phosphide wafer production line to China.
Later, Beijing Tongmei gradually took over all of AXT's physical business operations, including the production of indium phosphide wafers, while its North American parent company only handles capital operations. In other words, all indium phosphide wafers produced by AXT are theoretically made in China, and even if shipped back to the U.S., they must go through China's export procedures for goods.
These three links, whether through natural endowment or acquired advantages, have allowed China to seize the initiative in the indium industry chain. But the more lethal blow is the heavy hand of export controls.
In February last year, China included indium phosphide and related products in the export control list, and the first to be hit was Beijing Tongmei (AXT). Although officials stated that it obtained the export license in June, the "quantity was less than expected," resulting in lower-than-expected revenue in the fourth quarter of that year.
The long-term uncertainty brought by export controls has placed an even heavier shackle on the already fragile supply chain. AXT stated in its second-quarter earnings call that even with strong demand and sufficient capacity, its ability to ship to customers outside of China depends on the licensing decisions of the Chinese government [5].
But that's not all.
By the beginning of this year, China extended its controls to the more upstream indium, capping the exportable volume at 30% of total production, and further tightened the export approval for high-purity indium of 6N grade and above [4], directly holding a knife to the main artery of indium phosphide.
Japan's Sumitomo and JX Metals have become the targets of criticism. Over 90% of their high-purity indium raw materials are purchased from China. Reports indicate that due to supply constraints, their annual capacity has dropped from 180,000 pieces to less than 100,000 pieces, and the delivery cycle has been extended from less than half a year to 18 months.
Under such circumstances, one can understand why Coherent's CEO was so anxious to come to China to build a temporary friendship and plead to "spare the goods."
But the ones deciding whether Coherent and others can ship the goods are not just the Chinese government, but also the Chinese manufacturers in the chain.
Up to the Task?
Undoubtedly, under the dual pressure of supply falling short of demand and tightening controls, the biggest beneficiaries are China's indium phosphide industry chain.
To be fair, in the chain of "primary indium -> high-purity indium -> indium phosphide wafer," the high ground truly occupied by China is the most upstream primary indium.
In terms of high-purity indium, Chinese manufacturers mainly have a capacity advantage in mid-to-low-end products of 6N grade and below. High-end products of 6N/7N grade that can be used in optical chips are still dominated by Japanese companies such as DOWA and JX Metal. Although several Chinese enterprises have made breakthroughs in recent years (such as Zhuzhou Keneng), the road to validation is long.
The situation for the Chinese team in indium phosphide wafers (excluding Beijing Tongmei) is similar, namely strong in the mid-to-low end and weak in the high end.
The low-end 2-inch wafers have the highest degree of domestic production. The mid-end 3/4-inch wafers can already compete with overseas rivals, capturing most of the domestic market. However, on the most advanced 6-inch wafers, the domestic production rate is less than 5%, and counting all sizes, the global share is less than 10% [4].
The current export controls have brought the most direct benefit of order transfer and growth.
As overseas giants face obstacles in obtaining raw materials from China, supply uncertainty has surged. Downstream customers in urgent need of goods will naturally lean towards more stable Chinese manufacturers, especially Chinese local customers who have the advantage of proximity.
Yunnan Germanium is a domestic leader in indium phosphide wafers and currently the only domestic manufacturer capable of mass-producing 6-inch indium phosphide wafers. In 2025, its domestic operating revenue increased significantly by 60.77% year-on-year (reaching approximately 1.016 billion RMB). According to its performance forecast for the first half of this year, its non-recurring net profit surged by over 500%.
Its subsidiary, Yunnan Xinyao, signed an indium phosphide wafer supply agreement with a customer last month, with a total amount as high as 570 million to 855 million RMB (tax inclusive), which is even higher than Yunnan Germanium's total revenue for the entire year of 2025.
Further upstream, the high-purity indium giant Zhuzhou Keneng is no less impressive. In 2025, its domestic order revenue grew by over 60%, with domestic revenue breaking through the 1 billion RMB mark for the first time. Meanwhile, its overseas revenue skyrocketed from less than 30 million RMB in 2023 to nearly 200 million RMB in 2025.
Getting high-end products to the table is another benefit gained by domestic manufacturers.
In Yunnan Germanium's expanded indium phosphide wafer capacity and production plan this year, 6-inch products are explicitly included, proving that its large-size products have crossed from the laboratory R&D stage into the validation sprint and delivery stage.
Zhuzhou Keneng not only stably supplies 6N/7N grade and above high-purity indium to overseas indium phosphide giants like Vepics, but its ultra-high purity 8N grade, representing the industry's top level, has also broken the monopoly of the U.S., Japan, and Europe, and has achieved substantive sales.
At this point, the only two problems left for the domestic indium phosphide supply chain are validation and capacity.
The further upstream you go in the chip industry, the stricter the customer validation. The validation cycle for indium phosphide wafers usually takes several months to a year, and for high-purity indium, it is even longer, typically 1-5 years. Before establishing cooperation, downstream customers will first conduct multiple, small-batch product validations.
Currently, the manufacturers that have made breakthroughs in the high-end field are not just Zhuzhou Keneng and Yunnan Germanium, but a large portion of them are stuck at the validation stage.
For example, Yuguang Gold & Lead has achieved stable mass production of 7N high-purity indium. It is rumored that it has already sent samples to leading enterprises such as Huawei and Accelink Technologies for validation. According to its latest response on the investor interaction platform, it is currently in the stage of downstream customer sample certification, waiting for feedback results from downstream customers.
According to institutional estimates, the global demand for indium phosphide wafers will reach 2.6 million to 3 million pieces in 2026, but the global effective capacity is only 600,000 to 700,000 pieces, with a gap rate as high as 70% [7]. In terms of high-purity indium, according to industry media data, the indium gap in the optical communication field in 2026 will be about 50 tons [8].
As Nvidia's CPO accelerates mass production and high-speed optical transceivers rapidly ramp up, this gap is highly likely to be further expanded, becoming the decisive battleground between Chinese manufacturers and overseas giants.
By the end of 2025, Yunnan Germanium's indium phosphide wafer capacity was 150,000 pieces/year, with a planned capacity increase to 180,000 pieces in 2026. Meanwhile, Japan's Sumitomo has a monthly capacity of 39,000 pieces, estimating an annual capacity of over 400,000 pieces. The gap remains obvious.
Yunnan Germanium's under-construction "High-Quality Indium Phosphide Single Wafer Construction Project" is said to add 300,000 pieces of annual capacity. Japanese rivals are also aggressively increasing capacity; it all depends on who can land it faster.
By the end of 2025, Zhuzhou Keneng's annual capacity for "5N and above" (including 6N/7N grade) high-purity indium was 150 tons/year. Its new IPO fundraising project, the "Annual Production of 500 Tons of Semiconductor High-Purity Materials and Recycling Project," will add 60 tons/year of high-purity indium capacity. Japanese giants like Dowa have not announced their expansion plans.
A report from the U.S. investment firm Rosenblatt Securities in May this year revealed that Nvidia has urged suppliers to increase the capacity of indium phosphide lasers by 20 times before 2030, but ultimately only received a guarantee of a 12-fold increase [9].
What to do about the remaining 8 times? No more words, just bow your head and work hard.
Epilogue
In 2011, domestic financial media published a news report stating that the trading price of metal indium "surged" from4.6 RMB per gram to6.2 RMB within a month.
This was unprecedented in the domestic metal indium market. Over the past decade, 80% of the world's metal indium was exported from Yunnan, but the domestic trading price of metal indium remained as steady as an electrocardiogram. The ones pulling the strings behind the scenes were the Japanese.
Japan entered the indium industry in the 1980s but suffered from barren land that could not grow indium metal. Although China held most of the indium reserves, its capacity was scattered, manufacturers were small in scale, and bargaining power was poor, so Japanese manufacturers sniffed out the opportunity.
The Japanese bought indium at low prices all over China, then processed it into components in Japan, and sold it back to Chinese companies at dozens of times the price [3].
This is an early scar on the indium industry chain, and another wake-up call for China's industrial manufacturing:
Downstream always holds higher pricing power and more generous profit margins. Mastering upstream resources does not mean you can sit back and relax; instead, you should use this as an advantage to fully develop processing and manufacturing capabilities.
In 2011, the Yunnan Pan-Asia Non-ferrous Metals Exchange was established. By integrating supply, centralizing quotations, and allowing institutional and individual investors to participate in trading, it took the bargaining power back into the hands of Chinese manufacturers.
This was the starting point for China's indium industry to regain its voice. Today's optical transceiver hurricane is another beginning.
References
[1] Lumentum CEO warns of impending bottleneck on critical material used for silicon photonics — fab and material shortfall already lags 30% below customer needs as co-packaged optics demand skyrockets, tom's HARDWARE
[2] Distribution and production of indium resources, Asian Metal
[3] Who sets the price of indium? 21st Century Business Herald
[4] Significant supply-demand mismatch in indium phosphide, focus on opportunities for domestic substitution, Guosheng Securities
[5] AXT's Premium Multiple Prices an InP Recovery Under China's Export Control Cloud, kavout
[6] Financial reports of Yunnan Germanium and Zhuzhou Keneng
[7] Gap exceeds 70%! Indium phosphide concept surges again, multiple listed companies issue urgent reminders, 21st Century Business Herald
[8] China's CPO Industry: Shortage levels and choke point ratings of 9 key materials, China Powder Network
[9] Stocks In AI Bottleneck Soar 1,000%. Now Nvidia Is Demanding 20X More Supply, 247wallst