China is not beating the U.S. in the artificial intelligence race, at least not yet. But the country’s AI products are cheaper and good enough that a number of American AI users are shifting away from homegrown products. Enter a Silicon Valley panic. The U.S. government may take action, and a new Chinese model — Moonshot AI’s Kimi K3 — may be the trip wire.
‘Something good enough’
Kimi K3 can “allegedly beat some of the best systems built by U.S. companies at a fraction of the cost,” said The Verge. It is the latest in a series of cheap Chinese-made “open-weight” AI models that “anyone can download, inspect, modify and run on their own infrastructure,” said Microsoft. This presents a different approach from closed-source “proprietary platforms” like Gemini, Claude and ChatGPT; the Chinese method challenges the conventional wisdom that “leading-edge AI requires heavy spending on data centers and the advanced chips” used by U.S. companies, said CNN.
The Chinese open-weight models are “increasingly almost as good as closed systems made by leading U.S. companies,” which is why a “growing number of Americans” are moving to the Chinese systems, said The Associated Press. Kimi K3 “just seems snappier” than Anthropic’s Claude Fable chatbot, Mozilla chief technology officer Raffi Krikorian said to the outlet. Other Chinese open-weight models lack the “overall, full-range capabilities” of the products made by American companies, said the AP. But AI users “just don’t need” expensive homegrown AI products,” technology executive Curt Meinhold told the outlet. “We need something good enough.”
‘Full AI communism’
Some American AI executives are “sounding the alarm” about the rise of open-weight models, said The Wall Street Journal. The Chinese models represent “full AI communism,” in which governments provide artificial intelligence as “digital public infrastructure” rather than a “market product,” OpenAI’s Dean Ball said on X. The White House may soon decide to “create large amounts of regulatory risk around the use of open-weight Chinese models” as a result.
Others disagree. Restricting access to open-weight models “wouldn’t stop their proliferation — but would weaken U.S. startups” that use the technology, a coalition of 200 smaller tech companies argued last week in a letter to administration officials, per Politico. Regulation would be “great for Anthropic. We’re all going to have to spend money on Anthropic,” Suhail Doshi, the founder of AI infrastructure startup Particle, said to the outlet.
The White House is “showing signs it could ban cutting-edge Chinese AI models,” said Axios. Administration officials last year considered an executive order mandating that U.S. firms could “only host Chinese models if they could guarantee security and take liability if it were breached.” The ascendance of Kimi K3 is “reigniting” those old regulatory efforts. An example: The White House this week banned Chinese-made “humanoid and quadruped robots” to “protect the U.S. AI build-out,” said Reuters.