So, President Donald Trump's State of the Union address didn't mention critical minerals. That's fine—speeches are for the big, shiny topics. But behind the scenes, his administration is reportedly cooking up what could be one of the most interesting industrial policy experiments in a long time. And it involves letting an artificial intelligence system, built by the Pentagon, start telling people what things should cost.

According to people familiar with the matter, Trump intends to deploy this AI system to set prices for critical minerals. Think of it as a high-stakes pricing referee, but one that runs on algorithms instead of a whistle.

"The administration is still, in good faith, trying to respond to industry demand signals by creating an architecture of reliable investment, but it doesn't have the one tool that everybody kind of wanted them to use," said Eric Robinson, a consultant and former managing director of the Pentagon's Office of Strategic Capital. That missing tool, apparently, is a clear price signal—and now they're building an AI to provide it.

How Do You Teach an AI About Fair Prices?

The whole effort revolves around something called the Open Price Exploration for National Security (OPEN) program. It's an AI-driven metals pricing model that was launched by the Defense Advanced Research Projects Agency (DARPA) back in 2023. You know, the same folks who helped invent the internet.

Here's how it's supposed to work: The AI doesn't just look at what minerals are trading for today. It tries to estimate what they should cost by factoring in everything from labor and processing to logistics. Crucially, it also tries to strip out the effects of what U.S. officials see as market distortions. So instead of reacting to short-term price swings, this AI would help establish long-term reference prices—a kind of North Star for what things ought to cost.

This approach isn't happening in a vacuum. It's meant to anchor a new trade framework between the U.S. and its allies, an idea that Vice President JD Vance and others have been pushing. The vision is a club where everyone agrees to use these AI-calculated prices as a baseline.

Starting Small, Thinking Big

The initial focus won't be on massive markets like copper or lithium. Instead, they're looking at niche but strategically vital metals: germanium, gallium, antimony, and tungsten. These are the unsung heroes of semiconductors, defense systems, and advanced manufacturing. Their markets are relatively small and thinly traded, which makes them particularly susceptible to influence—especially from Chinese supply, which dominates production for many of these materials.

Beijing has long said its export policies play by global trade rules. But U.S. officials have a different view: they believe that concentrated production has artificially depressed prices, which in turn has discouraged Western companies from investing in their own mining and processing capabilities. It's a classic chicken-and-egg problem—prices are too low to justify new mines, but prices are low because there's no alternative supply.

The AI price model is supposed to break that cycle. If adopted by an allied trade bloc, it could serve as a reference point for contracts between miners and manufacturers. The idea is to reduce the uncertainty that comes with opaque, volatile markets. If you're a company thinking about building a $500 million processing plant, you'd really like to know what you can expect to sell your output for in five years. An AI-generated reference price might give you more confidence than today's spot market.

The Tariff Stick

Here's where it gets really interesting. The plan represents a clear shift away from the direct subsidies or price guarantees that often characterize industrial policy. Instead, the enforcement mechanism would be tariffs.

Imagine the AI decides that germanium should cost $X per kilogram. If imports start coming in priced below that reference level, they could face duties. The goal isn't necessarily to formally impose a price floor, but to use the threat of tariffs to push market prices toward the AI-calculated benchmark. It's a way of policing prices without the government directly setting them.

Of course, no system is perfect. Nathaniel Horadam, the former U.S. Department of Energy Critical Minerals Lead, pointed out that there are no guarantees here. Even with a reference price, multiple producers would still compete on price. And for this to have any meaningful impact on trade flows, the U.S. would need to persuade dozens of countries to join the system—no small diplomatic feat.

Then there are the practical questions: How often would the AI adjust its prices? Would the same prices apply uniformly across all participating nations? These aren't minor details; they're the kind of things that can make or break an international agreement.

Meanwhile, the market is already paying attention. The VanEck Rare Earth and Strategic Metals ETF (REMX) is up 23.33% year-to-date. That's not just random noise—it's investors positioning themselves for a world where critical minerals policy is getting a lot more interesting, and a lot more high-tech.