China's 70% Indium Phosphide Grip Tightens Two-Front Squeeze on U.S. AI Buildout

For AI companies racing to scale data centers, the physical materials underpinning compute power are often an afterthought. But with China controlling 70% of the indium phosphide needed for high-speed optical interconnects, the AI industry faces a supply chain bottleneck that could slow the very hardware it depends on. The simultaneous information warfare—allegedly stirring public opposition to data centers—adds a softer but equally strategic layer to this geopolitical competition.

America’s ambitious AI infrastructure build-out is facing a two-pronged challenge from China, one that strikes at both the physical supply chain and the information ecosystem. In early May 2026, Coherent Corp. CEO Jim Anderson revealed to investors that his company was running critically low on indium phosphide, a specialized compound that converts electricity into laser light—essential for the high-speed optical interconnects in data centers. The shortage was not a random market fluctuation; it stemmed directly from China’s tightened export controls over the material, which the U.S. Geological Survey says originates from China to the tune of about 70% of global production. Within a week, Anderson found himself on a plane to China alongside President Donald Trump, scrambling to secure the export licenses his factories needed, according to Reuters.

“

But with China controlling 70% of the indium phosphide needed for high-speed optical interconnects, the AI industry faces a supply chain bottleneck that could slow the very hardware it depends on.

This episode encapsulates the material front of the squeeze. Indium phosphide is a linchpin in modern AI computing clusters, enabling the rapid data transfer between chips that underpins training runs for large language models. Without it, data center operators cannot scale as planned. The Chinese Communist Party (CCP), keenly aware of this dependency, has over the past year steadily restricted exports of indium phosphide and other critical minerals. The result: rising prices and elongated delivery timelines for U.S.-based chipmakers and data center providers. Experts note that while these restrictions have not yet crippled U.S. AI expansion, they have significantly raised the cost and complexity of the build-out, forcing companies to seek alternative sources or redesign components—efforts that take years.

But the squeeze has a second, less tangible dimension. According to OpenAI and several U.S. lawmakers, operatives aligned with China have been fueling public opposition to AI data centers in America. The alleged campaign sows disinformation about the environmental and grid impacts of these facilities, aiming to slow permitting and spark local resistance. Officials say the effort is largely coordinated, leveraging social media and grassroots groups to erode support for new projects. While the immediate impact is hard to quantify, this information front compounds the material delays: a project that finally secures rare minerals might still be held up by community lawsuits or regulatory hurdles, extending the overall timeline.

Stephen Xia, a former engineer with the People’s Liberation Army who now tracks AI developments, describes the competition as an integrated contest that spans algorithms, data, talent, computing power, and—critically—supply chains. “If the CCP chokes off rare minerals,” he told The Epoch Times, “it would at the very least slow the pace at which the United States expands its high-end chips and data centers.” This viewpoint underscores a strategic design: China, trailing in AI capabilities, is using its natural resource dominance to impose friction, thereby buying time for its own domestic AI industry to catch up.