Washington May Push Cheap Chinese AI Out of the US

The cheap Chinese AI models that have been dragging prices down all year may not stay easy to use in the United States. Axios reported on Monday that the Trump administration is weighing several ways to push models like Moonshot's Kimi K3 out of American companies, and the launch of Kimi last week is what reignited the effort.

There is no signed ban and no executive order yet. What there is, according to sources close to the administration, is a set of levers being dusted off, and a real fight inside the White House over whether to pull them.

What is actually on the table

The measures under discussion are indirect by design. Nobody is proposing to outlaw a downloadable model, partly because open weights make that close to unenforceable. You cannot recall a file that is already on thousands of hard drives.

Instead, the reported options work by pressure. Commerce considered adding Chinese AI labs to its Entity List, which would cut off US access without a special license. The White House looked at an executive order that would let US companies host Chinese models only if they could guarantee security and accept liability for a breach. The NSA and the national cyber director's office weighed a formal advisory warning companies off Chinese labs entirely.

One source summed up the approach: "What's actually happening is slower and more durable," citing procurement rules, Entity List threats, and public pressure campaigns aimed at the US companies using these models.

Why Kimi K3 set this off

All of these ideas were floated last year and killed by officials who worried about strangling innovation. Two things changed.

The people who blocked the earlier efforts, including former White House adviser Sriram Krishnan, have left, and national security hawks have grown louder. And Chinese models kept getting better and cheaper. American companies increasingly reach for open models from China because they cost a fraction of the domestic options and, as Kimi K3 showed, are nearly as capable.

That combination, a stronger product plus a friendlier internal audience, is why the ban talk is back.

The fight inside the White House

The most striking part is that the loudest objection came from inside the administration. David Sacks, an outside White House AI adviser, wrote on X that "the leading closed labs, already a duopoly in terms of AI model revenue, want the government to eliminate their open-source competition."

Sacks has long warned about regulatory capture, the pattern where the biggest incumbents get rules written that lock out smaller rivals. A source told Axios that leading labs or their allies approach the administration every three to five months with a fresh idea for banning open-source models.

The counterargument is not fake. Chinese models do raise real questions about data handling and potential backdoors, and the government worrying about that is reasonable. The catch is that the American labs raising those concerns also happen to charge many times more per use, so their security worry and their business interest point in the same direction.

Why it matters to you

David Sacks, a White House AI adviser, put the tension bluntly: the biggest US labs want the government to knock out their cheaper open-source competition.

Both things can be true at once. Chinese models raise genuine security questions, and the American labs sounding the alarm charge roughly ten times more per use.

Those cheap open models are a big reason AI prices keep falling. Squeeze them out of the US market and the downward pressure on what everyone pays goes with them.

This idea has been floated and killed several times already, so I am not hitting the panic button. It is worth watching with a raised eyebrow, though, since the people who stand to gain the most are the ones already charging you the most for AI.

Sources

← Back to AI for Mortals