China‘s tightening of its indium phosphide export controls is sending shockwaves through the AI industry. The country has quietly ramped up scrutiny on shipments of this critical material, which is essential for the high-speed optical chips that power modern data centers. These restrictions took effect in early 2025, and they threaten to disrupt the supply chain for a component that has become indispensable to artificial intelligence infrastructure. If you follow the semiconductor or AI hardware sectors, this is a development you’ll want to understand — it could directly affect how quickly AI models can be trained and deployed in the coming years.
Why Indium Phosphide Is Critical for AI Data Centres
That’s because indium phosphide is the material powering the high-speed connections inside AI data centres. When you train or run large AI models, you’re not just relying on a single accelerator — you’re using clusters of GPUs or other AI chips that need to talk to each other at incredible speeds. The optical interconnects that link these accelerators depend on components made from indium phosphide. Without those components, the data transfer bottleneck would throttle performance, making it impossible to run the largest models efficiently. This explains why indium phosphide export controls are such a big deal for the AI industry: they directly affect the hardware that makes modern AI possible.
The Role of InP in AI Optical Interconnects
In an AI data centre, every accelerator must exchange massive amounts of data with its neighbours. This happens through transceivers and optical components that convert electrical signals into light and back again. Indium phosphide optical chips are the heart of these transceivers, enabling data rates that silicon simply can’t match. Because InP can both generate and modulate light efficiently on a single chip, it keeps the optical links compact, power-efficient, and fast enough to keep up with the demands of parallel AI workloads. When you see a rack of AI accelerators, the invisible network behind them — the cables and optics — is built largely around indium phosphide.
Why No Ready Substitute Exists
You might wonder whether other materials could step in. In photonics — the science of using light for data transmission — indium phosphide has no ready substitute. Silicon photonics can handle some tasks, but it struggles to generate light directly and typically requires separate laser sources. Other compound semiconductors like gallium arsenide work at lower speeds or less efficiently at the wavelengths used in data centre optics. For the combination of high speed, direct light generation, and integration density that AI clusters need, indium phosphide remains the practical choice. That’s why any restriction on its supply — such as the new Chinese checks — sends ripples through the entire AI hardware chain, from chip designers to data centre operators.
China’s Dominance in Indium Production and the New Export Controls
Given how crucial indium phosphide (InP) has become for AI hardware, the source of that material matters a lot. China’s near-monopoly on indium production gives its export controls outsized leverage over global supply chains. You might not think about where raw materials come from when you use AI-powered tools, but this concentration creates a single point of failure for the entire industry.
China’s Indium Monopoly
China produces around 70% of the world’s indium. That’s a staggering share for any critical material, especially one that sits at the heart of high-speed optical networking and AI chip manufacturing. When the country tightens checks on indium phosphide export controls, the rest of the world pays attention — because there simply isn’t enough supply elsewhere to fill the gap quickly.
Export controls on InP took effect in early 2025. The new rules mean anyone outside China must apply for special licenses to buy indium phosphide substrates or wafers from Chinese suppliers. In practice, this creates delays, uncertainty, and higher costs for companies that depend on a steady flow of this material.
Comparison with Previous Export Controls
This isn’t Beijing’s first move in the critical-materials game. Beijing has previously deployed controls on gallium, germanium, and rare earths. Those earlier rare earth export restrictions China introduced caused major jitters in the electronics supply chain, driving prices up and forcing manufacturers to scramble for alternative sources. The pattern is familiar: China identifies a material where it holds a dominant position, then uses that leverage to influence global markets or geopolitics.
The Chinese export controls gallium germanium experience offers a useful parallel. When those restrictions hit, companies that hadn’t diversified their supply chains faced production delays. The same scenario is now playing out with indium phosphide, only this time the stakes are higher because of AI’s insatiable demand for bandwidth and speed. Understanding your China indium production share exposure — and whether your suppliers are affected — is becoming a practical business concern, not just a geopolitical talking point.
Price Surge and Supply Chain Disruptions
This concern is now translating into stark numbers that affect your bottom line. The price of indium phosphide wafers has more than tripled since the controls began, straining AI infrastructure buildout. A six-inch InP wafer has climbed from roughly $1,400 to about $5,000, making every component significantly more expensive. For companies dependent on these wafers, this indium phosphide export controls policy directly increases costs for AI accelerators and networking equipment.
Wafer Price Surge
This InP wafer price increase isn’t just a statistic; it’s a practical hurdle. If you’re scaling AI systems, the higher prices may force budget adjustments or timeline extensions. The cost jump reflects growing demand for bandwidth and tightening supply due to regulatory checks. Understanding these dynamics is key to navigating the current market, and staying ahead requires close monitoring of price trends.
Coherent’s Warning and Licensing Delays
Nvidia-backed chipmaker Coherent warned of a shortage earlier this year, underscoring the urgency of the situation. Coherent’s chief executive personally raised the licensing delays indium phosphide during a US business delegation’s visit to China, highlighting how bureaucratic hurdles are affecting supply lines. This Coherent InP shortage warning is a clear indicator that you should pay close attention to regulatory timelines. Licensing delays can stall shipments, so planning ahead with buffer stocks or alternative suppliers might help mitigate risks. The combination of soaring prices and red tape creates a challenging environment for anyone involved in AI hardware procurement.
Global Production Capacity and the Challenge of Building Alternatives
Given those procurement hurdles, you might wonder how quickly alternative sources can step up to fill any gaps. The reality is sobering: building indium phosphide production capacity outside China is a multi-year, capital-intensive endeavor with no quick fixes. Even the companies already in the game face steep barriers to scaling up.
Current Non-Chinese InP Capacity
Right now, the global landscape for indium phosphide substrates is narrow. The largest players are concentrated in Asia, with AXT standing out as the world’s second-largest InP substrate producer. That position puts AXT in a critical role for any buyer looking to diversify away from Chinese supply. But being second-largest does not mean limitless output; AXT’s capacity is still modest compared to overall global demand, and its facilities are not positioned to absorb a sudden surge in orders overnight. Other non-Chinese producers exist, but their combined volume is far from sufficient to replace Chinese capacity. So when you evaluate your supply chain, the immediate takeaway is that InP production capacity outside China is stretched thin and not ready to compensate for sudden export restrictions.
Timelines and Investment Requirements
Expanding or building new InP substrate manufacturing investment is not a plug-and-play project. Setting up crystal growth, wafer slicing, polishing, and quality control lines requires specialized equipment, cleanroom facilities, and highly skilled technicians. You are looking at years of planning, permitting, construction, and ramp-up before a new fab reaches commercial volumes. Industry estimates suggest that even an aggressive timeline for a new facility can stretch beyond three to five years from groundbreaking to first shipments. That kind of capital outlay—ranging into the hundreds of millions of dollars—is only viable for companies with deep pockets and long-term confidence in AI-driven demand. For the procurement manager or hardware engineer reading this, the practical implication is clear: indium phosphide export controls are not a temporary blip you can ride out by quickly switching suppliers. They demand a strategic rethink of your sourcing plan, with an eye on years, not quarters.
Impact on AI Infrastructure and Affected Companies
If you are building out AI infrastructure, the InP shortage is a problem you cannot ignore. The reason is simple: indium phosphide goes into the transceivers and optical components that connect accelerators in AI clusters. Without those parts, your data centre becomes a collection of powerful chips that cannot talk to each other efficiently.
Some of the biggest names in the industry are already feeling the pressure. Coherent, a chipmaker backed by Nvidia, warned of a shortage earlier this year. That is a significant red flag. When a company closely tied to the world’s leading AI hardware supplier flags a supply issue, it suggests a bottleneck that could ripple across the entire ecosystem.
Another key player to watch is AXT, the world’s second-largest producer of InP substrates. If export controls tighten its ability to ship, the knock-on effect hits every company that relies on those substrates to manufacture their optical components. You might not buy directly from AXT, but your suppliers likely do.
So, how might the InP shortage slow down AI infrastructure buildout? Think of it as a domino effect. AI clusters need high-speed optical transceivers to move data between GPUs and memory. Those transceivers depend on InP-based lasers and photodiodes. If the raw material becomes scarce or expensive, production slows. Delivery times stretch. Costs rise. And the grand plans for expanding AI data centres get pushed back.
For companies already racing to deploy AI at scale, this is not a minor inconvenience. It is a strategic obstacle that requires you to rethink your supply chain resilience, not just your next order.
Comparing InP Controls to Previous Chinese Export Restrictions
This is not the first time China has leveraged its dominance in critical materials. Understanding how previous export controls unfolded gives you a clearer picture of what might come next for indium phosphide. Beijing has previously deployed controls on gallium, germanium, and rare earths, materials essential for semiconductors, fiber optics, and consumer electronics. These actions have consistently disrupted global supply chains and forced companies to rethink their sourcing strategies. The indium phosphide export controls follow a similar pattern, but the stakes are higher because of AI hardware demand.
Lessons from Gallium and Germanium Controls
When China introduced gallium germanium export controls, the immediate effect was price volatility and supply shortages. Manufacturers in the US, Europe, and Japan scrambled to secure alternative sources or stockpile inventory. The episode taught the tech industry a hard lesson about over-reliance on a single supplier. The indium phosphide export controls are likely to provoke a similar response. Indium phosphide is critical for photonics and high-speed electronics in AI systems, so any disruption will directly affect production timelines and costs. Companies that survived the previous controls understand the importance of diversifying suppliers now.
Government Responses and Trade Tensions
The US response to Chinese export controls has typically combined diplomatic pressure with incentives for domestic production. For example, rare earth export restrictions impact prompted Washington to fund research into recycling and alternative materials. With indium phosphide, the reaction is likely to be more aggressive because of its role in AI infrastructure. Governments in the US, Europe, and Asia are already reviewing their exposure to Chinese supplies, and trade tensions are rising. For you, the pattern is consistent: these controls create bottlenecks and force strategic adjustments. Watching how other governments respond will help you anticipate changes in availability and pricing.
Alternatives and Recycling Efforts for Indium Phosphide
With supply chains tightening, you might wonder what happens if indium phosphide becomes harder to get. The reality is that for photonics — the technology behind fiber optics and high-speed data transmission — there is no ready substitute. That makes the search for alternatives and better recycling practices more urgent than ever.
Research into Alternative Materials
Scientists are actively exploring other materials that could potentially replace indium phosphide in certain applications. For example, gallium nitride and silicon photonics are being studied for their ability to handle similar tasks, though each comes with trade-offs. Gallium nitride works well for high-power devices but struggles with the specific light wavelengths that InP handles naturally. Silicon photonics, meanwhile, benefits from existing manufacturing infrastructure but is less efficient for lasers and detectors. This InP substitute research is still in early stages, meaning a full replacement is likely years away, if it arrives at all.
Indium Recycling Initiatives
Because indium is a rare element, recycling efforts have gained attention as a practical way to reduce dependency on new mining. Indium recycling initiatives focus on recovering the material from discarded electronics, solar panels, and manufacturing waste. The process is complex, as indium is often used in tiny amounts within layered components. However, companies and research labs are developing better methods to extract it efficiently. If these indium recycling efforts scale up, they could help stabilize supply without relying solely on new exports. For you, this means keeping an eye on recycling infrastructure could be just as important as watching trade policy.
Frequently Asked Questions
What are the challenges of building InP production capacity outside China?
You face high capital costs and a need for specialized fabrication expertise when setting up indium phosphide production lines outside China. Securing a consistent supply of ultra-pure raw materials and qualified engineers adds further complexity. It typically takes several years to reach the quality and yield levels required for commercial semiconductor applications.
How do these controls compare to previous Chinese export restrictions on gallium and germanium?
The indium phosphide export controls follow a similar pattern to earlier restrictions on gallium and germanium, targeting materials essential for advanced technologies. The main difference is that indium phosphide directly impacts high-speed optical components used in AI data centres, making its supply chain more tightly linked to the AI boom. Both sets of controls aim to manage access to critical inputs while encouraging domestic processing.
Why is indium phosphide so critical for AI data centres?
Indium phosphide is a key material for manufacturing high-speed lasers and photonic integrated circuits that handle massive data transfers in AI infrastructure. These components enable the optical interconnects you need to move data quickly between servers during model training and inference. Without indium phosphide, achieving the bandwidth and energy efficiency required for large-scale AI workloads becomes extremely difficult.
