President Donald Trump’s latest comments on semiconductor exports sounded almost conciliatory — until they weren’t. Speaking recently on “60 Minutes,” the president said he would let Nvidia “deal with China” but drew a bright red line: Beijing could buy chips, just not the “most advanced” ones. The message was calibrated for maximum effect: permissive enough to please markets, hawkish enough to claim toughness. Nvidia’s stock jumped immediately — but China did not get what it wanted.

Days later, in a Financial Times interview, Nvidia’s CEO, Jensen Huang, warned that if the U.S. blocked his company from selling more of its advanced chips to China, it would “lose” the AI race. The argument was astonishing in its candor: Cut us off, Beijing wins.

As grateful as America should be for breathtaking innovations, an irreconcilable tension exists between national interest and fiduciary duty.

The comparison between a president sounding measured and a CEO trying to sound indispensable captures a dangerous inversion of power. Nvidia has become more than America’s most valuable company. It’s attempting to become its policymaker, shaping the boundaries of what Washington thinks possible in its competition with China.

To understand how one company reached that position, it helps to revisit what happened in Washington just days before Trump met Xi Jinping in South Korea.

Nvidia called it a GPU Technology Conference. Yet the event felt less like a developer’s conference and more like a tech-bro-meets-MAGA jamboree: free swag and a booming video hymn to American genius — from Thomas Edison to Donald J. Trump. Huang, leather jacket gleaming, strode out like a preacher to proclaim that the age of reindustrialization had arrived.

The D.C. version of GTC was not the San Jose GTC tech insiders have come to know. For the first time, Nvidia brought a full-blown edition of its developers’ confab to the capital, a strategic choice. The company does not merely want to sit at the table where policy is made — it wants to own it.

After hours of Super Bowl-style buildup — financiers whispering, tech CEOs hinting — attendees were herded into a dimly lit hall, where Huang unveiled a cascade of partnerships. The headline act that made sleeves roll up on both the policy bench and the brokerage floor was the Vera Rubin Superchip, billed as made in America and spoken of with the gravity reserved for national monuments.

It’s a dazzling feat of engineering: silicon that can be waved before a crowd as proof that America can still design, assemble, and scale. Expected to debut next year, the chip is music to policy wonks’ ears, a gleaming symbol of reindustrialization, and perhaps a psychological hedge against the fragility of Taiwan. For investors, it’s manna. As robots increasingly take charge, building chips in the U.S. will keep the supply chain close to home and safeguard companies against the whims of geopolitics.

Then, with the applause fading, an undercurrent of tension lingered, one that perhaps only the wonks could fully register. After that opening montage, capped by Jensen’s almost rhetorical question, “Was that video amazing?” the subtext became harder to ignore. And when he closed his remarks by thanking the audience “for your service and for making America great again,” it was impossible not to think of what the financiers were murmuring on the next stage over.

“Nvidia will — or should — ship more GPUs to China.” “Jensen’s flying straight to Korea after GTC to meet Trump.” “A deal’s coming.”

Those were among the refrains traded by figures like Cantor Fitzgerald’s C.J. Muse and Altimeter Capital’s Brad Gerstner. All this, of course, is contrary to the prevailing consensus among China-watchers that the notion of rendering Beijing dependent on Nvidia’s chips is fantasy. Cultivating indigenous capability by acquiring American technology by legal or illicit means has long been Beijing’s modus operandi.

Huang knows this. Still, his company has long worked to blunt export controls and push China-specific versions of its flagship Blackwell chip, the so-called B20. It’s a familiar playbook: First came the H100, then its “export-compliant” cousins, the H800 and H20. Each time, Washington tightens the rules; each time, Nvidia finds a workaround. But this must stop.

RELATED: Big Tech’s AI boom hits voters hard — and Democrats pounce

Photo by Ron Jenkins/Getty Images

The dilemma is simple but corrosive. As grateful as America should be for breathtaking innovations, an irreconcilable tension exists between national interest and fiduciary duty. Huang may sound bullish on “betting on America,” but the reality is starker: If his company could power the AI revolutions of both superpowers at once, it would add trillions to its market cap. He is pragmatic and coldly arithmetic. Build the best chips, profit from ubiquity. You don’t get where he is without knowing your math.

At GTC, I saw the divide play out in miniature. As Altimeter’s Brad Gerstner floated the idea that “logic is on the side of letting Nvidia compete with China,” I turned to a biotech researcher. Blunt and unamused, he said: “Bulls**t.” He went on to explain that, in his field especially, China’s ascent has been a wholesale rejection of the “make China dependent” fantasy. He wasn’t wrong: Under Xi Jinping, the Made in China 2025 agenda has rendered such dependency theories delusional.

Huang tries to thread the needle gracefully, extolling U.S. manufacturing while signaling an embrace of Chinese developers. As an American, it’s hard not to be charmed by his all-American chip. As a realist, however, one leaves with questions no press release can answer. In a way, the release of this patriot-approved superchip was meant to suggest, “See, now we can sell some Blackwells to China.” As charmed as one can be, the answer is still no.

One could have told the Roosevelt administration that cutting Germany off from nuclear materials would stifle innovation. Yet we did exactly that during the Manhattan Project. And we won. It may not sound like it, but this is the same choice we face today — only this race has even greater implications for the future of civilization.

The goal can’t be attempting to trap Beijing in “dependency.” The stakes are too high. The most prudent approach is to focus on surpassing them in innovation while closing loopholes that let Beijing do what it has mastered: Learn from us, then try to replace us.

Jensen Huang has every right to fight for his company’s profits. But foreign policy shouldn’t run on a corporate playbook. The U.S. needs innovators — not influencers — setting the terms of technological rivalry.

Editor’s note: A version of article appeared originally at the American Mind.