Chinese Tech Giants Circumvent U.S. Export Restrictions by Training AI Models Overseas

In response to stringent U.S. export restrictions on advanced semiconductors, leading Chinese technology firms, including Alibaba and ByteDance, have begun training their artificial intelligence (AI) models in overseas data centers, particularly in Southeast Asia. This strategic move enables them to access Nvidia's high-performance chips, essential for developing large language models, without directly violating U.S. sanctions.

The U.S. government's April 2025 decision to impose export licensing requirements on Nvidia's H20 and AMD's MI308 AI chips to China has compelled Chinese tech companies to seek alternative avenues for AI development. By leasing data center space from non-Chinese entities abroad, these firms maintain access to critical computing resources necessary for advancing their AI capabilities.

An exception to this trend is DeepSeek, a Chinese AI firm that stockpiled Nvidia chips prior to the export bans and continues to train models domestically. DeepSeek is also collaborating with local chipmakers, including those led by Huawei, to develop the next generation of AI training hardware.

The U.S. export restrictions are designed to prevent China from advancing its AI capabilities, which are seen as a potential threat to national security. By training AI models overseas, Chinese companies are navigating these restrictions, raising questions about the effectiveness of such measures. This development also underscores the global nature of technology supply chains and the challenges in enforcing unilateral export controls.

This is not the first instance of the U.S. imposing export controls to limit China's technological advancements. Similar measures have been implemented in the past, targeting various sectors, including telecommunications and quantum computing. However, the current situation highlights the adaptability of Chinese firms in circumventing such restrictions by leveraging international resources.

The strategic shift by Chinese tech giants to train AI models overseas reflects the complex interplay between technological advancement and geopolitical strategies. It raises important questions about the efficacy of export controls and the future landscape of global AI development.