China's control over a critical material is creating a significant bottleneck in the global race for AI dominance.
At the heart of this issue is Indium Phosphide (InP), a compound semiconductor that serves as the foundation, or substrate, for high-speed lasers and photonic integrated circuits. Think of these as the superhighways for data within AI data centers. They are essential for moving massive amounts of information quickly between powerful AI accelerator chips, like GPUs. Without a steady supply of high-quality InP wafers, building the next generation of AI infrastructure becomes incredibly difficult and expensive.
The problem began in February 2025, when China, which produces about 70% of the world's primary indium, placed InP on its export control list. Since then, obtaining export licenses has become a slow and unpredictable process. This isn't just bureaucratic red tape; it's a calculated strategic move. Beijing is using its dominance over this upstream material as a geopolitical lever, directly impacting U.S. companies that rely on it.
This situation didn't arise in a vacuum. It's a clear 'tit-for-tat' response in the escalating tech rivalry between the U.S. and China. First, the U.S. imposed strict export controls on advanced AI chips to China to slow its technological progress. In response, China is now leveraging its control over essential raw materials. By deliberately slowing down InP exports, Beijing is targeting a critical chokepoint in the U.S. AI supply chain.
The market impact has been immediate and severe. The price for a single 6-inch InP wafer has skyrocketed by about 250% to roughly $5,000. This cost shock is rippling through the supply chain, affecting optical component makers like Coherent and Lumentum. While these companies are scrambling to increase their own InP production in the U.S., building new capacity is a complex process that takes two to three years.
This means that for the foreseeable future, the pace of AI infrastructure development may not be determined solely by the availability of advanced GPUs, but by the supply of these crucial optical components. China has effectively turned a materials dependency into a strategic throttle, reminding the world that the path to AI supremacy runs through a complex and fragile global supply chain.
- Indium Phosphide (InP): A semiconductor material used to make high-speed electronic and photonic devices, essential for fiber-optic communication in AI data centers.
- Photonic Integrated Circuit (PIC): A chip that uses light (photons) instead of electricity (electrons) to transmit and process information, enabling much faster data transfer.
- Substrate: The base material on which a semiconductor device, like a chip, is built. It acts as the physical foundation.