The U.S. government is investing $500 million in an artificial intelligence startup as part of an effort to reduce reliance on imported materials critical to semiconductor manufacturing.
According to the department, the award will accelerate the development and deployment of SandboxAQ’s AI-driven materials discovery platform, which is designed to address bottlenecks in the semiconductor supply chain by identifying alternatives to scarce or foreign-controlled materials.
Backed by Nvidia, SandboxAQ will focus on four areas where U.S. semiconductor manufacturing remains dependent on overseas supply chains: alternatives to PFAS “forever chemicals,” advanced catalysts, rare-earth-free magnets, and new battery chemistries for semiconductor facility backup power systems.
The initiative aligns with the Trump administration’s broader effort to reduce dependence on China for strategically important industrial inputs.
China remains the dominant force in the global rare-earth supply chain, accounting for roughly 70 percent of global mining output and more than 90 percent of refining and separation capacity. Rare-earth magnets are essential components in advanced chip-making equipment, powering the ultra-precise actuators, vacuum pumps, and wafer-positioning systems used throughout high-end semiconductor plants.
China also plays a dominant role in the global battery supply chain. Although it produces only a modest share of the world’s lithium and cobalt ore, Chinese companies control more than two-thirds of global processing capacity for both minerals, giving Beijing substantial influence over downstream battery manufacturing.
“This award will accelerate the discovery and innovation of critical materials and reduce our reliance on foreign-controlled materials,” Commerce Secretary Howard Lutnick said in a statement.
The funding comes from the CHIPS and Science Act of 2022, a $280 billion package aimed at boosting domestic high-end semiconductor manufacturing, reducing reliance on overseas supply chains, and countering China’s technological influence and competition.
In exchange for the funding, the Commerce Department will receive a minority, non-voting equity stake in SandboxAQ.
While SandboxAQ identifies itself as an AI company, its technology differs significantly from consumer-facing AI products such as chatbots.
Rather than developing Large Language Models trained on text and code, the company specializes in what it calls “Large Quantitative Models,” which are trained on scientific data and designed to solve problems in physics, chemistry, and materials science. Its models are used to screen millions of potential compounds in software before researchers start to test them in the lab.
SandboxAQ says the approach can dramatically shorten the timeline for materials discovery. The company has claimed that its catalyst-development platform, built using 13.5 million simulations run on Nvidia hardware, can screen candidate materials roughly 20,000 times faster than conventional methods.
The funding is aimed at research and development rather than manufacturing. None of the proposed replacement materials has yet reached commercial-scale production, but officials hope the investment will accelerate their path from the lab to industrial deployment.
IBM received approximately half of that funding to establish a new subsidiary, Anderson, which the company said will manufacture quantum-grade wafers engineered to meet the exceptionally stringent purity and structural requirements needed to run stable quantum computing systems.
As with the SandboxAQ agreement, the Commerce Department took minority, non-controlling equity stakes in each of those companies, saying the arrangement is meant to “enhance the return for the U.S. taxpayer” while supporting strategically important technologies.
