US blacklist threat for 3 Chinese AI labs risks higher costs for firms

The US government is weighing trade blacklist designations for three Chinese AI labs — Moonshot AI, DeepSeek and Alibaba's Qwen team — as surging adoption of their cheaper open-weight models by American companies triggers national security concerns, according to people familiar with the matter.

"We are at a critical inflection point in AI policy. The leading closed labs, already a duopoly in terms of AI model revenue, want the government to eliminate their open-source competition," David Sacks, outside White House AI adviser, said in a social media post.

Chinese open-weight models price their application programming interfaces at a fraction of US rivals. DeepSeek-V4-Pro charges $0.87 per million output tokens, compared with $50 for Anthropic's Claude Fable 5 — a discount of more than 98 percent. Coinbase Chief Executive Officer Brian Armstrong said the exchange runs models including GLM-5.2 and Kimi in production, cutting overall AI spending nearly in half even as token consumption surged. The measures under consideration include placing the companies on the Commerce Department's Entity List, issuing a joint advisory from the National Security Agency and the Office of the National Cyber Director, and drafting an executive order that would hold US companies liable for security breaches involving hosted Chinese models.

A ban would disrupt a rapidly growing practice: US enterprises downloading open-weight models from repositories such as Hugging Face and running them on private servers, slashing inference costs while keeping proprietary data in-house. The current average US tariff on Chinese goods stands at about 19 percent after the 2025 escalation, and the last round of technology export controls reduced bilateral semiconductor trade by an estimated $23 billion over 12 months, according to Census Bureau data. If implemented, the blacklist could push US companies back toward higher-cost domestic alternatives from OpenAI and Anthropic, while Beijing is likely to retaliate with its own restrictions on critical minerals or rare earth exports.

The push marks a revival of efforts that stalled earlier this year after internal administration disagreements over market impact. Officials had previously considered adding DeepSeek to the Entity List, a trade blacklist maintained by the Bureau of Industry and Security that restricts foreign entities from purchasing American hardware, software and technology. Those measures were paused after pushback from free-market advocates including former White House adviser Sriram Krishnan, who argued the restrictions would stifle innovation and hand a monopoly to leading US labs.

The new urgency follows the July launch of Kimi K3 by Moonshot AI and Alibaba's Qwen 3.8 Max, both of which delivered benchmark performance competitive with frontier US models. In some tests, Kimi K3 matched or exceeded the capabilities of leading closed-source systems, narrowing the gap that US labs had relied on to justify premium pricing. The open-weight nature of these models — allowing users to download, customize and fine-tune them with corporate data — has driven adoption among US enterprises seeking data privacy and cost control.

Enforcement of any ban would face significant technical hurdles. Open-weight models exist as downloadable files mirrored across public repositories and independent torrents, making them difficult to recall once released. Companies routinely fine-tune, quantize or distill these models, blending the Chinese base with domestic corporate data until provenance blurs. Even under strict download bans, firms could host the models through foreign subsidiaries, though that vector runs into know-your-customer rules at cloud providers and the extraterritorial reach of US export controls.

The administration's strategy may not require an outright ban. According to the Axios report, government sources said procurement rules, Entity List threats and public pressure campaigns aimed at companies using Chinese models could achieve the same result. Officials also plan to highlight potential backdoors and security vulnerabilities in Chinese models, raising governance concerns that could deter enterprise adoption voluntarily.

This article is for informational purposes only and does not constitute investment advice.