When Donald Trump reacted to China’s export restrictions on rare earth minerals—a group of 17 chemical elements used in almost all electronics—by threatening a new 100 percent tariff on Chinese goods, Wall Street investors who had yawned at most of his erratic announcements for six months finally took notice. It was the same stock-tanking pressure that led Trump to climb down from his Liberation Day tariffs; sure enough, by Sunday, the TACO (Trump always chickens out) vibes started kicking in.
“I have a great relationship with President Xi … he’s a great leader for their country, and I think we’ll get it set,” Trump told reporters on Air Force One on the way to Israel. “Don’t worry about China, it will all be fine!” he exclaimed on Truth Social. Next thing you know he’ll be apologizing in Mandarin like John Cena.
U.S. officials later said that the ultimate outcome depended on China. But markets had bounced back by Monday, reasonably surmising that Trump is beholden to what he sees as the ultimate barometer of success—the stock market.
Holding the president of the United States on a tight leash is certainly a lucrative asset. But investors should probably be more wary of the situation. America has made an unusually directional economic bet that is at this moment totally dependent on Chinese rare earth exports. The circumstances that brought us here long predate Trump and are rooted in decades-long failures to retain our technological know-how and channel it into industrial production. It’s never too late for a wake-up call, but the country is in a terribly vulnerable position where China can snap its fingers and snuff out the only thing propping up our economy.
The bet I’m talking about, of course, is on artificial intelligence, and the astronomical buildout of infrastructure, mainly data centers. Close to half of the gain in gross domestic product this year will come from data center construction, and around 80 percent of stock market gains are attributable to a handful of AI-heavy tech companies. OpenAI announced a trillion dollars in computing deals this year like it was nothing, and we’re on track for $500 billion in annual AI capex spending by next year.
AI may be eventually worth that investment, but the revenue right now is paltry. The money is either coming as a cross-subsidy from dominant tech platforms looking to control the next big thing, or—even worse—rampant financial engineering. Herb Greenberg, a longtime financial columnist and stock analyst, told us on the podcast I co-host, Organized Money, that companies like Nvidia are “round-tripping,” funding the companies that buy their GPU chips (essentially paying for their products themselves), and creating venture capital arms to finance the sector. Some of this financing is coming from cash flow, but not all of it; there’s a ton of debt involved, financed mostly by private credit, which operates outside of the regulatory perimeter.
Incredibly, a signature villain of the 2008 financial crisis, the hedge fund Magnetar, is back as a top financier of CoreWeave, a data center operator that is losing money yet also funding other startups. The AI bubble is sometimes associated with the dot-com overbuild in the late 1990s, but the financing looks a lot like the run-up to the housing bubble, with big piles of debt-fueled garbage.
How does a bubble like this pop? It’s impossible to know. It looked for a minute like China’s release of DeepSeek showed that massive capex wasn’t necessary to advance AI tech, and Nvidia started losing altitude. But then everyone forgot about that. It’s clear that the valuations of several major companies are dodgy, their promises are empty, and the revenues needed for their profitability are nearly impossible. If investors decide not to be captivated by the notion of inventing God in a machine and cut their losses, the whole thing could crumble.
But there’s a lot of unwillingness to walk away from the God-machine dream. So a likelier end to the bubble could be external. And that’s where China’s rare earth mineral dominance comes in. China controls 70 percent of the mining, 90 percent of the processing, and 93 percent of the magnet manufacturing involving rare earths. It is a global chokepoint for electric vehicles, military weaponry, and virtually everything that needs a transistor or semiconductor. And that includes what’s needed to install and run data centers.
There’s a lot of unwillingness to walk away from the God-machine dream.
China has been pulling back on rare earths since Liberation Day, with softer embargos leading to slowdowns at U.S. factories. But last week they expanded these restrictions for 12 of the 17 types of rare earth elements. Licenses will be required for their export starting November 8, with a specific case-by-case review for developing advanced computing and memory chips. This comes weeks before a key trade meeting with the U.S., and was obviously announced for maximum leverage to roll back every sanction the U.S. has placed on China in recent years.
China is insisting it’s not banning rare earth exports. But it’s likely to make it much harder for U.S. AI companies to get them. And let’s be clear: denying access to rare earths would stop the AI buildout dead. Given how much that buildout is responsible for masking recessionary winds and juicing aggregate growth, it would really stop the U.S. economy in its tracks, too. The U.S. needs China’s critical inputs far more than China needs U.S. customers, which makes this threat credible.
That we got to such a point is totally insane. In my 2020 book Monopolized, I explained that the U.S. invented rare earth magnets for fighter jets in the 1970s. “A company called Magnequench manufactured these magnets in Indiana until 2004, when the plant closed and shipped out to China,” I wrote. “In fact, China bought the entire company, using as a front a hedge fund operated by Archibald Cox Jr., son of the famed Watergate prosecutor.”
It took decades to hollow out industrial production, hand over critical inputs to other countries to monopolize, and then base our economy on development that needs those critical inputs. No single president is responsible for creating this vulnerability; it was a team effort of short-sightedness. But it’s true, as Rush Doshi has said, that Trump bumbled into a trade war “without preparation, without allies, and without reducing our own vulnerabilities.”
Rare earths aren’t that rare; it’s the processing where China really excels. And we’ve had a decade-and-a-half of knowing that China was capable of wielding rare earths as a geopolitical weapon. The country briefly cut off Japan in 2010 after a skirmish involving disputed islands in the South China Sea. Nine years later, Xi Jinping pointedly toured a rare earth factory in the middle of Trump’s first trade jostling with China.
U.S. efforts to diversify the rare earth supply chain have been honestly pathetic. There have been a lot of developments of blueprints of plans, and more recently small funds for production. Trump’s Defense Department took a stake in MP Materials, which runs the biggest rare earth mine in America. But none of this represents the kind of serious resources and market commitments that are necessary.
U.S. efforts to diversify the rare earth supply chain have been honestly pathetic.
Alex Jacquez, who handled rare earth policy in the Biden administration, also came on Organized Money, and explained that rare earths are a fairly low-margin product, and investors see no upside in getting behind domestic projects. “It’s hard to justify to shareholders why you would make a big capex investment in a big magnet facility that’s not going to pay off for a couple years,” he said. Besides, we don’t make electronics here either, so it’s easy for China to take the rare earths they create, pass them to manufacturing plants also inside their country, and maintain the near-monopoly. Once you abandon a technology, it’s hard to get it back.
Trump has one primary go-to move, so he threatened tariffs, and China threatened right back. At the same time, Trump worked to wind down the renewable energy and electric vehicle production in America that gave rare earths a self-reinforcing domestic market. “If we are going to do industrial policy, we have to do it. There is no halfway in, halfway out,” Jacquez said.
Ultimately, this explains Trump’s make-nice change of heart. He can’t afford to do anything else, because we need rare earths. But it was obvious China was going to take out its big leverage hammer. Trump has repeatedly claimed he was blindsided by China’s threat. How? You can’t just conduct a trade war and expect no pushback, especially from a country with many options to do so.
The cynic’s take would be that by denying or limiting rare earths, China is doing the U.S. a favor by effectively popping a bubble before it gets even bigger. Whether AI is a useful technology is immaterial to the mania of overbuilding the infrastructure to realize it and paying for it through dodgy accounting. It may be better to put a stop to all that now, even if it’d be painful.
But China is also revealing something vital about our country. Development and abandonment of critical technologies has been a consistent pattern in recent U.S. history, across solar panels and batteries and semiconductors and pharmaceuticals. We have been committed to nothing so much as not producing anything. And now we’re at the mercy of outside forces for our economic fortunes.
Maybe we can go back to the drawing board, develop magnets that don’t involve rare earth minerals, and refrain from making the same stupid mistakes again by retaining the technology. As a general principle, we have the power to fix this, though maybe not with the current self-destructive leadership. But for now, we’re destined to drive up to an endless series of tollbooths, forced to pay up and play by someone else’s rules. That is the result of conscious decisions.
