中/EN
Frontline Hotspot
Frontline Hotspot

Musk and Huang "Back China" on Same Day: Chinese AI Shifts from Follower to Peer Competitor

Musk says China will likely become the AI leader; Huang says China is destined to produce excellent AI technology. Hugging Face data shows Chinese model downloads at 17.1% surpass US 15.8% for the first time. China considers AI export controls—role reversal.

Published July 27, 20264 min read

1. The Event: Two Giants "Back China" on the Same Day

On July 25, two bombshells dropped almost simultaneously:

Elon Musk stated bluntly in a 90-minute Economist interview: "At some point in the future, China is very likely to become the AI leader. Even if the US bans Chinese models, it won't stop them."

Jensen Huang said in a foreign media interview: "Outstanding talent will always find outstanding answers. China is destined to produce excellent AI technology. We should continue learning from them and cooperating with them."

This wasn't Huang's first such statement. On July 24, he had already said: "China's open-source AI models are excellent. The US should not fear China's open-source AI models."

Three statements in two days—an unusually dense frequency.

2. Background: Hugging Face Data Speaks

Why the sudden collective "pro-China" stance? Look at the data:

On AI model download platform Hugging Face, from August 2024 to August 2025, Chinese model downloads accounted for 17.1% of total downloads, surpassing the US's 15.8%.

This is the first time Chinese AI models have overtaken the US on the world's largest open-source platform.

More importantly, Chinese models aren't just "being downloaded"—they're "being used." Models like Qwen3, DeepSeek, and GLM achieve inference costs one-tenth of GPT-4o while matching or exceeding performance.

3. Musk's "Timeline"

In this interview, Musk laid out an aggressive timeline:

  • Within 5 years: AI intelligence surpasses the sum of all human intelligence
  • After 10 years: Money loses its value, human work is fully replaced

Musk, who once firmly called for "pausing AI," has now completely reversed his stance: "The AI wave is unstoppable. Even if there were an off switch, I would never press it."

Why? Because he's done the math: the productivity gains from AI will usher humanity into an "era of universal abundance." The question isn't "whether to have AI," but "how to distribute AI dividends."

4. Role Reversal: China Considers AI Export Controls

At the same time as Musk and Huang's statements, another development is telling:

China is considering expanding export controls to cover advanced AI models and certain semiconductor technologies.

The Ministry of Commerce has already held discussions with Alibaba, ByteDance, Zhipu AI, and others, covering:

  • How to restrict core training data transfers overseas
  • How to limit foreign users from downloading model weights
  • How to prevent key chip design technology outflows

The tables have completely turned. Previously, the US was choking China (chip bans); now China is considering choking the US (AI model export controls).

5. Implications for Ordinary People

  1. Don't bet on "who wins": US-China AI competition isn't a zero-sum game. Both sides will produce top-tier models. For developers, "usable, good, cheap" is what matters.

  2. Open source is the moat: China's models overtook not through closed-source monopoly, but through open-source ecosystems. Qwen, DeepSeek, GLM are all open-source—global developers can use and modify them.

  3. Cost is the killer app: DeepSeek's inference cost is one-tenth of GPT-4o. This isn't "cheap means bad"—it's real efficiency gains from architectural innovation (MoE, quantization, distillation).

In one sentence: Musk and Huang's statements aren't "polite words"—they're "acknowledgment." Chinese AI has shifted from "follower" to "peer competitor," and in some areas has begun "leading."

References:

This article is AI-assisted and human-edited. Last updated: 2026-07-27

FAQ

Why did Chinese AI models surpass the US on Hugging Face?
Mainly through open-source ecosystems and cost advantages. Qwen, DeepSeek, and GLM are all open-source; inference costs are one-tenth of GPT-4o while matching performance.
Will China's AI export controls affect ordinary developers?
Not in the short term. It's still in discussion phase, mainly targeting large-scale transfers of core training data and model weights. Ordinary developers using open-source models are unaffected.
Is Musk's "5-year AI surpasses all humanity" claim credible?
It's an aggressive prediction, not consensus. Most researchers believe AGI requires more time. But AI surpassing average human performance in specific domains (coding, writing, analysis) is already fact.

Related