Congress should share, not shield, US artificial intelligence tech

China recently made a major move in its ongoing tech race against the U.S. The Chinese unveiled the fastest supercomputer in human history, called LineShine.

Before this week, the U.S. had held that honor, thanks to its El Capitan machine at Lawrence Livermore National Laboratory in California. China’s victory is more symbolic than anything else, but it does raise an important question: How was China able to pull this off despite years of increasingly stringent U.S. export restrictions?

The U.S. imposes significant restrictions on tech exports to China, including on graphics processing units, as well as the machines used to manufacture them. These units provide supercomputers with the enormous processing power they need to function.

Yet China’s LineShine record-holder doesn’t use graphics processing units. It effectively worked around American trade barriers by relying on a massive network of standard microprocessors.

Therein lies a lesson for the U.S. Washington is trying to hobble the Chinese tech industry by blocking it from importing U.S.-made semiconductors and other products. Yet as LineShine shows, these restrictions haven’t always worked; oftentimes, they’ve just forced China to adapt.

The same point could be seen earlier this year when the federal government reversed course on whether Nvidia could export its H20 chips to be used in sophisticated Chinese AI models.

The Trump administration commendably recognized that if the Chinese were going to develop AI, they ought to at least use American chips.

Unfortunately, Beijing told the Americans: “thanks but no thanks.” In the face of........

© The Hill