Meta CEO Mark Zuckerberg said the United States should develop a strategy to compete with China’s artificial intelligence sector rather than impose a ban on it. In a recent interview, he argued that improving technology, not restrictions, is the way to preserve the U.S. position as a leader.

The debate has moved beyond questions about how users will access AI chatbots. It now centers on who will shape the next generation of AI platforms, developer ecosystems and infrastructure. That matters because the firms that set those standards can influence how AI tools are built into software, business workflows and consumer products. The United States remains the leader in the sector, but the rapid progress of Chinese models has outpaced what policymakers had planned.

American models still dominate global usage

A RAND Corporation study of AI-driven internet traffic across 135 countries found that large language models from the United States accounted for 93% of worldwide internet traffic in August 2025.

The market has also expanded rapidly. The number of users visiting top AI companies rose from about 2.4 billion in April 2024 to around 8.2 billion in August 2025. U.S. platforms continued to grow even as competition from China increased.

U.S. hardware dominance also supports its software lead. According to Epoch AI’s AI chip sales data, Nvidia’s H100-equivalent accelerators still dominate frontier AI computing capacity worldwide. That is one reason Washington continues to focus on semiconductor export controls, even as Chinese AI software becomes more popular.

DeepSeek expanded China’s global reach

The most significant shift came after DeepSeek launched its R1 reasoning model in January 2025. RAND said China’s share of worldwide AI traffic rose from nearly 3% to almost 13% in just two months, while traffic to China’s AI platforms increased by around 460%.

Chinese companies achieved more than 10% market share in 30 countries and captured at least 20% of AI traffic in 11 markets, especially in developing economies and in countries that have close ties with China.

RAND said cost is one driver of the growth. Chinese systems typically cost between one-sixth and one-fourth as much as comparable American models, while improvements in multilingual performance have weakened another traditional U.S. advantage.

Epoch AI’s frontier model repository shows that more companies can now build cutting-edge foundation models. While many of the strongest projects still come from U.S. labs, Chinese organizations are responsible for an increasing share of frontier model launches.

RAND’s research also found that China has invested heavily in AI diplomacy, but concluded that businesses, developers and users, rather than government efforts, are primarily driving adoption. That makes the competition harder to contain with policy alone, because usage patterns can spread through commercial and developer demand.

Export controls remain Washington’s other lever

A ban on Chinese AI models would add to existing U.S. restrictions on advanced AI chips, although the effectiveness of such measures has been mixed.

The Center for Strategic and International Studies (CSIS) reported that the Trump administration loosened restrictions on shipments of Nvidia’s H200 accelerators to China in December 2025 and moved to a case-by-case review of requests to export H200 and AMD MI325X chips.

CSIS said China has blocked local companies from obtaining the chips even when U.S. licenses were approved, while promoting the importance of domestic alternatives. The episode reflects a more complicated reality: export controls may reduce China’s access to leading-edge hardware, but they can also strengthen local semiconductor production and the use of generative AI tools that are difficult to restrict once released.

Zuckerberg and Washington differ on AI strategy

Zuckerberg’s view runs counter to the prevailing line in Washington.

“The U.S. is the AI leader in the world. We’re an AI superpower. China is second.” — U.S. Treasury Secretary Scott Bessent

Bessent’s position is echoed by Interior Secretary Doug Burgum.

“The U.S. cannot lose to China in the AI arms race.” — Interior Secretary Doug Burgum

Brookings has argued that treating AI as a simple race misses key distinctions. The U.S. advantage comes from large-scale commercial activity, private investment and venture capital-backed technological progress, while China’s approach relies on heavy industrial policy and building a self-reliant technology ecosystem.

Brookings also said the most important advantage may not be chatbot market share itself, but the surrounding developer ecosystem. Consumer preferences can change quickly, but companies are less likely to switch AI systems once they are embedded in workflows and applications.