US is reportedly pressuring dozens of allies to clearly pick sides in the US-China AI race — an escalation of the chip export ban that began in late 2022, moving from "stranglehold" to "bloc formation."

What this is

Since the US banned Nvidia from exporting top-tier AI chips to China in late 2022, this "tech decoupling" has been quietly advancing. The latest move: Washington is no longer satisfied with point sanctions — it now demands that allies take a position at the institutional level on whether to join the US-led AI export control bloc.

Three categories of countries most affected:

  • Chip manufacturing nations: Netherlands (ASML), Japan (Tokyo Electron), South Korea (Samsung, SK Hynix)
  • Compute deployment nations: UAE, Saudi Arabia — these Middle Eastern countries were originally key launchpads for Chinese AI companies going abroad
  • Model and cloud service nations: Southeast Asia, India — local data center site selection is being reassessed

The logic behind the move is clear: point restrictions can only delay China's AI progress by 6-12 months; systematically pulling allies into the same bloc is required to extend that time window.

Industry view

Supporters (mostly the US position): AI is next-generation military and economic infrastructure, and letting high-end compute flow to strategic adversaries is "unilateral surrender." Former US national security advisors and other officials have publicly backed this logic.

Opponents (what we find more worth watching): Many European and Asian countries view this as being "dragged onto a war chariot." Netherlands' ASML has repeatedly stated that excessive controls will damage its own interests; Japan's industry has also warned that AI supply chains are already deeply intertwined, and a hard cut will backfire on US semiconductor companies' R&D investment.

A calmer assessment: This "enclosure-style" policy may force out two independent AI ecosystems — one based on US chips + US models, one based on Chinese chips + Chinese models. In the short term, the US still has a 2-3 year lead; in the long term, the Chinese ecosystem's self-circulation capability will become stronger.

Impact on regular people

For enterprise IT: Companies operating in China need to reassess their AI compute procurement strategies; domestic alternatives (Huawei, Cambricon, Hygon) will see significantly elevated priority.

For individual careers: Chinese users of overseas AI tools (GPT, Claude, Gemini) may face more unstable service; at the same time, iteration speed for domestic AI tools (DeepSeek, Qwen, ERNIE) will be accelerated by strong demand.

For consumer markets: The globalization of AI products may be ending — future AI phones and AI assistants purchased in China will differ dramatically from US versions in capability, compliance, and content; the "AI experience" for cross-border travel and cross-border work will also split.