America’s self-defeating AI export controls

US AI policy helps China; needs to flood allies with tech, build AI bloc in America. (Illustration: WSJ)
US AI policy helps China; needs to flood allies with tech, build AI bloc in America. (Illustration: WSJ)

Summary

If it’s hard for allies to buy U.S. chips, they will get them from other suppliers—including the Chinese.

America’s response to China’s rise in semiconductors and artificial intelligence has been reactive, timid and structurally unserious. The U.S. is forfeiting a long-term strategic advantage.

This isn’t about consumer-facing chatbots or viral image generators. Artificial intelligence isn’t a tech trend. AI infrastructure is a foundational technology platform. Whoever is first in infrastructure, adoption and standards will define the next economic era and shape the future of commerce, war and governance.

In efforts to maintain America’s head start, U.S. policymakers are targeting graphics processing unit access and semiconductor tooling with a regime of ever-expanding export controls, with an emphasis on restricting Chinese GPU acquisition. That focus misses the point.

Exporting hardware isn’t the same as exporting capability. Cutting off China’s access to American chips might have slowed model development, but it also sharpened Beijing’s skill at diffusion in another way. China now leads the world in open-source AI, and that lead has emerged under constraints that U.S. policymakers imposed. What America blocks temporarily in outputs (models), it enables permanently in inputs (chips).

The same pattern already played out in telecoms with ZTE and Huawei semiconductors: Cheaper alternatives undermined U.S. dominance. America should know how this ends.

If AI is eating the world, we must ensure that it eats on American hardware and resources. The U.S. should be flooding the world with American GPUs in a concerted way. Untargeted export controls don’t stop adversaries, but they do punish allies and accelerate technical independence from American frameworks. The American strategy should be simple: Let every aligned country run on Nvidia silicon and American standards.

Today, Biden-era restrictions target countries such as Switzerland, India, Portugal, Israel and Argentina alongside such countries as Yemen, Kazakhstan and Algeria. Mexico, which assembles a significant portion of our GPU servers, requires U.S. approval to transfer or re-export these servers. That’s policy incoherence, not strategy.

The U.S. shouldn’t fear that our adversaries will use our technology as much as it should fear our allies will use our adversaries’ technology. Every U.S.-aligned deployment of AI infrastructure becomes a forward operating base. Giving ground to Chinese alternatives is giving up geopolitical leverage.

Reshoring shouldn’t stop at the U.S. border. It should extend across the Americas. The logic is familiar: a hemispheric manufacturing base, protected by U.S. security guarantees, reinforced by export markets—just as America built with the original Monroe Doctrine. The U.S. must treat the Western Hemisphere as a strategic manufacturing block. Leaders such as Nayib Bukele in El Salvador and Javier Milei in Argentina are signaling a generational shift in Latin American leadership—one open to deeper cooperation.

Nvidia has announced hundreds of billions in U.S.-based expansion. Apple is rethinking its supply chains. Original-equipment manufacturers are building components closer to home. But the U.S. shouldn’t confuse a national-security priority with an economic one. Security must be economically feasible. An “America+1" strategy, with Latin America as the natural extension of our manufacturing capacity, is both efficient and strategically sound.

Next, America should deepen the Indo-Pacific tech alliance. Japan, South Korea and Taiwan are more than trading partners. They’re our technical and strategic front line. Their complete alignment with American AI frameworks is essential not only to contain China but to serve as our ambassadors of infrastructure across the region.

This America+1 model achieves two goals: resilient supply chains and exclusive access to the world’s best AI tools and platforms. If America wants to reshore and near-shore AI infrastructure, it can’t build high walls. Tariffing the input goods needed to power our manufacturing renaissance undermines the rest of the effort.

Overly strict export controls shrink U.S. market share, raise costs and hand Chinese firms the opportunity to step in with cheaper alternatives. America is writing the go-to-market strategy for China’s semiconductor sector while punishing countries that can’t afford finished goods at scale. If the U.S. can buy raw materials for 10 cents and sell back final products for a dollar, that’s a win for America.

If China builds and exports a sovereign AI stack and then spreads that infrastructure into our alliances and trade routes, the U.S. will lose the ability to define global standards. With that comes the loss of control over the levers of power in the AI era. But if the U.S. becomes the default platform for AI infrastructure, America will preserve both technological leadership and economic leverage while extending U.S. power into the next generation.

The U.S. won the internet by being first and open. AI will work the same way. Either lean into the system that works—openness, innovation and export-led alignment—or lose ground to authoritarian mercantilism.

America wins when the world buys American. Let’s let them.

Mr. Ginn is CEO and co-founder of HydraHost, a venture-backed AI data-center services and management company.

Catch all the Business News, Market News, Breaking News Events and Latest News Updates on Live Mint. Download The Mint News App to get Daily Market Updates.
more

topics

Read Next Story footLogo