America’s Semiconductor Crossroads: PA’s Role in Securing Minerals for the AI Age
AI data centers are colossal buildings, often spanning more than one million square feet. Microsoft and OpenAI’s joint facility in Wisconsin will cover 1.2 million square feet; Meta’s Louisiana project exceeds four million square feet; and Google has two projects underway at 1.4 and two million square feet. But what truly matters is what is inside these concrete walls. Networks of advanced supercomputers fill the newly constructed space, supported by intricate cooling systems, backup power generation, gigantic batteries, and other essential infrastructure. At the core of all these systems are semiconductors.
Few realize that the Lehigh Valley was the original “Silicon Valley.” Western Electric’s Allentown manufacturing plant produced the first mass-manufactured transistors in 1951, paving the way for modern semiconductors – the foundation on which microchips are built. Though Bell Laboratories invented the first transistor in 1947 in New Jersey, it partnered with Western Electric to scale production. The Lehigh Valley’s access to markets, railways, waterways, world-class universities, and manufacturing expertise positioned it as a leader in this emerging field.
Transistors soon evolved into modern semiconductors, driving microchip innovation in the decades that followed. Pennsylvania and the Lehigh Valley remained central to this development and still retain some production capacity today. However, domestic westward expansion – fueled by inexpensive land, a favorable business climate, military spending, and engineering-focused institutions such as Stanford and UC Berkeley – shifted the industry’s epicenter to California, establishing the Silicon Valley we know today.
By 1970, the United States accounted for 48% of global semiconductor production, with Silicon Valley dominating worldwide research, design, and manufacturing. U.S. market share peaked in the early 1980s at about 55%. But in the late 1970s, the Japanese government recognized the industry’s importance and invested $300 million ($1.8B today) to launch a public-private partnership with its six leading computer companies. This marked the beginning of a dramatic global shift.
By 1989, Japan controlled 51% of the global semiconductor market, while the U.S. share dropped to 35%. Rather than competing head-on, American companies increasingly outsourced manufacturing to East Asia, seeking lower labor costs and looser regulations, while keeping research and development onshore. As Japan’s dominance waned, Taiwan, South Korea, and later China cut further into U.S. production. Through the 1990s and early 2000s, this trend accelerated. Today, the United States accounts for only about 10% of global fabrication capacity. Meanwhile, Japan, China, Taiwan, and South Korea together produce more than 75% of the world’s semiconductors. Most concerning is China’s rapid rise – from just 7% of global production in 2005 to a projected 25% by 2025.
China’s ramp-up in semiconductor manufacturing should come as no surprise. The most common semiconductor materials are silicon, germanium, gallium arsenide, and indium phosphide. In 2023, China produced 6.6 million tons of silicon. The next largest producer, Russia, produced just 620,000 tons, followed by Brazil at 390,000 tons. The United States did not even rank in the top 10. China also accounts for more than 60% of the world’s germanium, more than 80% of the world’s gallium, and more than 70% of the world’s indium. No other country comes close to China’s dominance in producing these critical minerals – the necessary inputs for modern semiconductor and microchip manufacturing.
One data center project requires hundreds of thousands of advanced microchips to achieve the massive computing power necessary for AI operation. Only a small fraction of all microchips are AI-capable, making this a major pinch point that could threaten future investment. Can global semiconductor and microchip supply keep up with this massive demand? More specifically, is the United States especially vulnerable given its lack of critical mineral mining and refining?
Not long ago, the U.S. learned this vulnerability the hard way. Between 2020 and 2022, auto production stalled – not due to shortages of steel, rubber, or labor, but because of a lack of semiconductors, microchips, and related imports. Today, demand is growing even faster: Sen. David McCormick recently announced $90 billion in AI investments in Pennsylvania, driving unprecedented demand for the minerals embedded in supercomputers, servers, and storage systems.
Bringing new mines and processing facilities online is time-consuming and costly, challenges the Trump administration has committed to solving through long-term investment. To confront this vulnerability, the president has issued executive orders on “Unleashing American Energy” and “Immediate Measures to Increase American Mineral Production,” building on his September 2020 order on supply chain awareness and resilience. These directives are reinforced by $1 billion in new grants just announced by the U.S. Department of Energy.
Collectively, these programs join other U.S. government actions to strengthen national security, energy independence, and industrial competitiveness. For Pennsylvania, these programs offer grants, matching funds, and other support, appealing to a state with significant advantages and an industry community already demonstrating excellence in the field.
Pennsylvania can be the keystone in America’s push for a secure critical minerals supply chain, leveraging its abundant, underutilized resources and established industrial infrastructure. Coupled with its historic role in the birth of the modern microchip, the Commonwealth is poised for a classic “voyage and return” story. Its industrial legacy and leadership in coal provide workforce and technology expertise to support project development. Feedstock potential exists in acid mine drainage and coal waste, making Pennsylvania a compelling proving ground. Successful demonstration projects could simultaneously remediate damaged land. For example, researchers at the University of Pittsburgh published a study finding 40% of the total lithium used in the U.S. could come from Marcellus Shale drilling wastewater. However, because the bulk of global lithium refining capacity exists overseas, primarily in China, our underlying national challenge remains.
But are the federal government’s efforts enough? Even if we access raw critical minerals, they must still be refined into usable forms. Again, China dominates. The People’s Republic of China currently refines 83% of the world’s copper, 73% of its lithium, 97% of cobalt, 98% of graphite, and 96% of rare earths (including essential inputs for semiconductors). A 2025 International Energy Agency report noted: “In December 2024, China restricted the export of gallium, germanium, and antimony—key minerals for semiconductor production—to the United States. This was followed by further announcements in early 2025, including restrictions on tungsten, tellurium, bismuth, indium, molybdenum, and seven heavy rare earth elements.”
What is urgently needed is a Manhattan Project-style initiative to expand mineral mining and refining capacity in the United States and among trusted allies – creating secure supply chains and distribution networks for semiconductor and microchip production.
Everything humans touch is grown from the earth or mined from it. While the information generated by AI companies is intangible, the machines behind it are not. Every component must be mined, refined, and manufactured to perform these modern tasks. If we fail to secure supply chains and distribution networks for critical minerals, we risk filling million-square-foot warehouses with nothing but empty space.
