US Nvidia chips to China: What does that mean for the AI race?
Loading...
During the Cold War, U.S. policy on export controls for advanced technology meant that the United States prevented sales to adversaries to stifle competition as rival countries vied to explore space and develop nuclear weapons.
So, while America鈥檚 leadership in technology has hinged on having a large and dynamic economy, it has also meant managing where its most advanced products are sold. As artificial intelligence emerged as the defining technology of the 21st century, and American tech firms created chips with the world鈥檚 highest computing power, the Biden administration aimed to keep the U.S. ahead by restricting sales, especially to countries such as China and Russia.
The U.S. gained a critical advantage in the AI race through its ability to manufacture advanced chips, small pieces of semiconducting material that form the backbone of technologies including AI computers.
Why We Wrote This
The Trump administration is allowing Nvidia to sell its advanced computer chips in China, a reversal of U.S. policy. The move is prompting concern that China could use the chips to close the gap or even win in the AI race.
But on Monday, President Donald Trump reversed years of U.S. policy by announcing that the government would allow a leading American tech company, Nvidia, to sell its second-most-advanced artificial intelligence chip to approved Chinese customers.
These H200 chips are far more powerful than the best China currently has. Mr. Trump鈥檚 decision has sparked debate about whether the U.S. is ceding its AI advantage to China, or whether, as the administration says, the deal will make China more dependent on U.S. chips and thus slow China's efforts to make its own chips that match or exceed those made in the U.S.
Mr. Trump said in a TruthSocial that 25% of the profits from the chip sales to China will be paid to the U.S. government, though he has not said how that will work. The president also said that the same approach will apply to other tech companies. The deal comes after Nvidia鈥檚 chief executive, Jensen Huang, spent months lobbying the White House to ease restrictions on exports.
Administration officials say the U.S. won鈥檛 be exporting its most advanced chips. But the stance appears to be shifting, since previously the U.S. had emphasized keeping as wide a gap as possible between its technology and that of other countries.
鈥淭his decision flies in the face of that,鈥 says Jacob Stokes, a senior fellow at the Center for a New American Security, an independent, bipartisan think tank based in Washington.
Opponents of the deal fear China will use American-made chips to learn how to build its own computer chips that can match the U.S. in computing scale. Many also accuse the Trump administration of handing away U.S. leverage for short-term profit.
At stake, experts say, is which country will win what鈥檚 essentially become an arms race to produce AI technology that could ultimately allow its owner to dominate global markets and gain military superiority.
A change in policy on export controls
The Trump administration鈥檚 deal has alarmed defense hawks 鈥 including both Republican and Democratic members of Congress, many of whom fear the technology sharing will help China modernize what is already the world鈥檚 largest standing army.
Beijing 鈥渨ill use these highly advanced chips to strengthen its military capabilities and totalitarian surveillance,鈥 read a by the House Select Committee on the Chinese Communist Party, chaired by Republican Rep. John Moolenaar of Michigan. 鈥淣vidia should be under no illusions. China will rip off its technology, mass produce it themselves, and seek to end Nvidia as a competitor.鈥
There is another view that aligns with the administration鈥檚 strategy. John Villasenor, a senior fellow at the center-left Brookings Institution鈥檚 Center for Technology Innovation, believes the deal will ultimately benefit the U.S. Expanding the advanced chips market to China will give American companies more money to pour into AI innovation, he says.
He also believes the deal will lead China to rely on U.S. chips instead of working to improve its own. If the U.S. government had continued to limit China鈥檚 access to critical chips, he says, that might have ended up forcing China to 鈥渟upercharge鈥 its own domestic investment into chip creation.
鈥淚t鈥檚 overly simplistic to think that by simply blocking China from accessing these chips, that鈥檚 going to impede China in the AI arms race,鈥 says Mr. Villasenor. 鈥淚t would have the opposite effect.鈥
Many experts point to the development of advanced semiconductor chips, such as the H200, as one of the few areas where the U.S. maintains a significant lead over China in AI development.
The deal won鈥檛 necessarily be a tipping factor in the arms race, says Mr. Stokes of the Center for a New American Security. But 鈥渋t is a significant concession that helps China deal with its biggest drawback or vulnerability on achieving advanced AI.鈥
China鈥檚 drive for self-sufficiency
China, however, might not embrace the American computer chips as some experts expect. The Financial Times on Tuesday that Chinese regulators have discussed instituting an approval process before companies could buy the imported chips, including giving a reason why domestic manufacturers weren鈥檛 meeting their needs. Beijing already requires this for a lower-performing Nvidia chip, the H20, that Mr. Trump permitted to be sold to China in August.
Still, Reuters on Wednesday that Chinese companies ByteDance and Alibaba have expressed interest in buying the newer, more powerful chips.
China has spent years working to reduce its dependence on the U.S. for AI development, producing products ranging from drones to chatbots. Aside from pouring money into domestic manufacturing, China has also sent messages to its own tech companies, discouraging them from using U.S. chips, particularly for government-related work.
鈥淐hina is moving as fast as possible to build its own chips and become self-reliant in chip production,鈥 says Mr. Stokes. 鈥淭he idea that this [deal] will keep China reliant on U.S. chip production 鈥 it appears to me that the ship has kind of sailed on that.鈥
But Mr. Villasenor doesn鈥檛 see it as an either-or scenario. China won鈥檛 stop investing in its own infrastructure to develop chips, he says, and they have made significant strides in building their own domestic capacity. But he says the country can鈥檛 afford to lose access to U.S. chips yet.
鈥淚t鈥檚 a long hill to climb,鈥 he says.