China controls 99 percent of the world’s primary gallium, a critical mineral and semiconductor crucial for building the microchips of the future. In 2023, it placed export controls on gallium to retaliate against American restrictions on the export of advanced chips to China. In December 2024, China escalated to an outright ban on gallium exports to the United States. The U.S. National Defense Stockpile had zero gallium reserves when that ban landed.
The United States has been here before. The United States pioneered and scaled modern silicon semiconductor infrastructure. A significant reliance on international manufacturing and the loss of domestic silicon dominance reflect a failure to recognize the importance of industrial capacity to national security. With silicon, the intellectual property was American, but the chips were “Made in Taiwan.” If similar blind spots persist, the United States risks repeating this failure with gallium nitride, a wide-bandgap semiconductor that outperforms silicon at high voltage, high frequency, and extreme temperatures. It’s the beating heart of every modern radar and electronic warfare system.
The answer to America’s vulnerability in gallium nitride-based chip manufacturing is not more fundamental research. What’s needed is a hard pivot to heterogeneous integration: a domestic supply chain — from gallium extraction and wafer fabrication to advanced packaging — built on dual-use, production-scale facilities that supply the warfighter and the American civilian at the same time.
Learning from Silicon
The United States solidified technical dominance in high-power, high-frequency systems through World War II-era efforts like the MIT Radiation Laboratory. That dominance was short-lived. Around the 1970s, semiconductor offshoring began. Intel was the first U.S. semiconductor company to offshore, but many followed. In 1987, Morris Chang founded Taiwan Semiconductor Manufacturing Company, built on a foundation of American knowledge. The U.S. Radio Corporation of America had trained Taiwanese engineers through a 1976 licensing agreement with Taiwan’s Industrial Technology Research Institute. As of 2024, Taiwan Semiconductor Manufacturing Company controls nearly 70 percent of the global foundry market.
Today, the United States retains design and intellectual property strength, but has little control over advanced manufacturing. The United States treated microchips as an economic commodity with few national security implications. The 2022 Creating Helpful Incentives to Produce Semiconductors and Science (CHIPS) Act was a response to these vulnerabilities, but the release of funds has been inexcusably slow.
The success of the act is debated. The National Center for the Advancement of Semiconductor Technology was shut down in August 2025. The act promised the creation of close to 50,000 jobs, yet Intel received approximately $7.8 billion in subsidies and still laid off around 35,500 workers in less than two years. GlobalFoundries received up to $1.5 billion and announced $500 million in share buybacks. Meanwhile, Taiwan Semiconductor Manufacturing Company’s Arizona fab has taken twice as long to build equivalent facilities to those in Taiwan and faced significant cost overruns, exposing the depth of the manufacturing gap the United States must close. The United States must learn from these mistakes with silicon. Repeating these failures with the next generation of semiconductors would be a colossal error of public policy and will.
Gallium Nitride: The Next Strategic Semiconductor
While silicon reigns supreme for digital and logic applications, the U.S. semiconductor revolution began with the advent of radar. Nearly a century later, revitalizing the domestic chip industry may depend on radar once again. Gallium nitride is the state-of-the-art for high-frequency, high-power amplifiers essential for modern radar. It handles high voltage, high frequency, and extreme temperatures simultaneously, the exact conditions of modern electronic warfare, making it the ideal semiconductor for radar systems as well as other national security applications.
Numerous U.S. ground-based and naval systems currently utilize gallium nitride. The Raytheon AN/SPY-6(V)1 is the Navy’s next-generation Integrated Air and Missile Defense radar. Lockheed Martin’s AN/SPY-7 powers allied Aegis Ballistic Missile Defense platforms. The Northrop Grumman AN/TPS-80 provides air surveillance and defense for the Marine Corps. Beyond sensing, gallium nitride is vital for electronic warfare. The Raytheon Next Generation Jammer Mid-Band enables the EA-18G Growler to disrupt adversarial communications.
Gallium nitride will not replace silicon. The two will coexist. The marriage of the two semiconductors is critical for the buildup of modern, highly optimized chips, an idea known as heterogeneous integration. Silicon can provide the digital backend for modern radar phased arrays, at scale. Gallium nitride enables high-frequency circuits with high-output power but lacks digital capabilities. Unless the United States is willing to build the advanced packaging facilities to do this at high volume, it will repeat the mistake it made with silicon chips.
Heterogeneous integration solves a problem that has constrained radar and electronic warfare hardware for decades. Gallium nitride provides unmatched radio frequency power. Silicon provides dense digital control. Piezoelectric films provide excellent filtering. Yet these materials cannot be grown on a single substrate. The industry assembles radio frequency front ends from discrete chips connected by traces that degrade signal quality, limiting performance, inflating cost, and preventing the scaling that next-generation phased arrays demand.
The challenge is not the science, but manufacturing at scale. The Next-Generation Microelectronics Manufacturing program at the University of Texas at Austin is a meaningful start, but focuses on prototyping, not production. The United States still lacks the high-volume advanced packaging infrastructure that next-generation defense systems will require, and Asia is building it faster.
The Gallium Problem
Primary gallium — gallium in its raw, unrefined form — is low purity. It must be refined further to produce high-purity gallium required for gallium nitride wafers. China produces 99 percent of the world’s primary gallium and has weaponized that position. In 2023, it placed export controls on gallium in response to U.S. chip restrictions, threatening the refined gallium supply on which U.S. gallium nitride foundries and allied wafer producers both depend on for in-house wafer growth and external wafer procurement. In December 2024, China escalated to an outright ban on U.S. exports. The ban was temporarily suspended in November 2025 following a U.S.-China trade truce, but it remains in effect for military end-users, and the suspension itself expires in November 2026. China can flip the switch again at any time.
The vulnerability exposed by the Chinese gallium export restrictions reveals a long-term industrial decline in U.S. gallium refinement. In 1985, the United States was the first country to have a dedicated gallium mine. The Apex Mine in southwestern Utah closed after just three years because international gallium imports produced as a byproduct of aluminum production were cheaper. Since then, the United States has had no domestic capability to produce primary gallium.
As of 2024, Indium Corporation operates the only high-purity gallium refinement facility in the United States, relying on imported semiconductor scraps. The Chinese government heavily subsidized aluminum production, allowing aluminum production to rise from 2.6 to 45 million tons between 2000 and 2025. As a byproduct, China’s gallium production capacity rose from 20 to 900 metric tons of output between 2000 and 2025.
Compounding this vulnerability, the U.S. National Defense Stockpile had no gallium reserves when China imposed its December 2024 ban. The Trump administration’s February 2026 launch of Project Vault — a $12 billion public-private initiative to stockpile critical minerals — is a step in the right direction, but even under optimistic projections, domestic production will only reach 10 to 15 percent of national consumption by 2030. China-based Innoscience, the largest 8-inch gallium nitride foundry in the world, leads in manufacturing capability. It also holds nearly 30 percent of the global gallium nitride power device market. This manufacturing capability is what matters in wartime.
Silicon and gallium nitride are distinct materials, but analyzing the policy and decision-making surrounding the two reveals striking parallels. The United States pioneered the silicon transistor, then watched domestic companies offshore and cede manufacturing to foreign competitors through failed intelligence oversight and ill-advised technology transfers. This gave rise to the foundry model and paved the way for Taiwan to dominate silicon manufacturing.
The gallium nitride story is following the same arc. The United States government pioneered gallium nitride through decades of government-funded research and maintains the majority of revenue in gallium nitride radio frequency devices, with U.S. companies dominating the defense and telecom segments. But the United States has never built the high-volume manufacturing infrastructure to match its design leadership. China has done so, through state-backed initiatives that mirror the government-driven industrial policies Japan and Taiwan used to dominate silicon. The gallium failure poses one critical distinction from the silicon blindside: In the latter case, silicon dominance was merely ceded to countries within the U.S. sphere of influence. A loss of gallium nitride excellence, however, hands victory to a direct strategic adversary.
Building a Heterogeneous Integration Supply Chain
The U.S. government must invest directly in national security-related technology companies and startups, with the taxpayer as the investor. The fabless mindset has led to an overfocus on designing chips and offshoring manufacturing. The United States needs an industrial revolution in heterogeneous integration: a standalone Department of Defense program authorized under Defense Production Act Title III, with four mandates. The Department of Defense’s newly established Economic Defense Unit — created to align defense strategy with economic competition and secure access to critical capabilities — could be the right vehicle to execute them.
First, establish a framework of standalone, dual-use advanced packaging production facilities. Each facility must be government-funded, industry-operated, and located near defense customers and cleared workforces. They should not be co-located with university research fabs. Academic fabs have few security boundaries, with foreign nationals roaming freely through them. Moreover, the competing demands of basic research distract from production. Sites near the Boston Route 128 corridor, Huntsville, Dallas-Fort Worth, and San Diego place production capabilities at the edge, adjacent to radar and electronic warfare customers who need it, supplementing facilities like MIT Lincoln Laboratory, rather than replacing them.
These facilities must be dual-use: The defense mandate builds them, while revenue from wireless, power electronics, and data center market sustains them. The Semiconductor Manufacturing Technology consortium of the 1980s excluded smaller companies and stagnated when government funding ended. The National Advanced Packaging Manufacturing Program piloting facility sits today with no funding or operator. These facilities must carry milestone-based production targets, not research outputs.
Second, establish a grant and equity program spanning the full heterogeneous integration supply chain, from gallium nitride wafer growth through advanced packaging. Startups will be the primary source of innovation in this space, as is the case now in many defense technologies. The government must inject startups at every stage of the supply chain through direct grants and tax credits. For production-scale facilities, the government should take equity stakes in critical companies, extending the precedent of the Trump administration’s $8.9 billion equity stake in Intel to companies for whom gallium is core. Qorvo and Wolfspeed are ideal candidates. The United States should view domestic champions not just as tickers on the stock market, but as key strategic proxies in a global competition for national security.
Third, for the fabrication of gallium nitride wafers, a secure source of domestic gallium is critical. Project Vault is a foundation to build upon. Building on the Department of Energy’s TRACE-Ga initiative and the Defense Production Act Title III award to ElementUS Minerals, the government must expand funding for novel gallium extraction techniques developed by startups. Strategic partnerships must establish a “Trusted Gallium Road” initiative, a network of allied nations ensuring a constant gallium supply outside the Chinese sphere of influence.
Fourth, the United States must think outside of the traditional semiconductor hubs. Advanced packaging does not require Silicon Valley types. It requires discipline, precision, and a workforce culture with deep industrial roots. The coal mining communities of Appalachia and the steelworkers of the Rust Belt, the manufacturing communities that built American industrial power but then watched it leave, are the same communities that can revitalize America again — this time in the cleanroom. A workforce development program modeled on the National Defense Education Act of 1958 would effectively create a talent pipeline in these communities. Government-backed advanced packaging facilities in communities uprooted due to workforce losses can supply the warfighter and ignite domestic stimulus.
Conclusion
The United States is at a strategic crossroads. The imperative is clear: Act now, or accept a slow and steady decline leading to semiconductor mediocrity. A loss of gallium nitride excellence means a win for China. The nation cannot afford to repeat the error it made by turning its back on the production of silicon chips. The moment has come for the United States to lead a revolution in chip manufacturing, starting with heterogeneous integration.
Pradyot Yadav is a National Defense Science and Engineering Graduate fellow at MIT’s Microsystem Technology Laboratories, with a minor in public policy and security studies of emerging technologies, working on 3D chips for next-generation radar. He has worked at Raytheon, Qorvo, and IBM, where he built the hardware this piece is about and other technologies core to national security. He has published several peer-reviewed publications with the IEEE. More about his work can be found at his website and LinkedIn.
**Please note, as a matter of house style, War on the Rocks will not use a different name for the U.S. Department of Defense until and unless the name is changed by statute by the U.S. Congress.
Image: Craig Fritz via Sandia.gov.
