The US Commerce Department wants to block remote access to advanced AI chips in China. This move shifts the battleground of US-China tech competition. The rivalry is moving from hardware sales to compute-as-a-service. This is not just about closing a loophole. It shows that the physical location of a chip matters less than who controls its power. Washington wants to make sure China cannot build a top-tier AI ecosystem using US-designed compute. This rule applies even if that compute sits in Singapore instead of Shenzhen.
This change moves beyond the first phase of export controls. Those initial rules focused on physical hardware. Now, the goal is to cut off access to performance. For Beijing, this makes the need to build domestic AI compute even more urgent. This rule turns a supply chain risk into a real bottleneck for Chinese cloud firms and AI developers. It forces them to rely on domestic options, even if those options are worse.
Korean media, like the AI Times, sees this as a natural but bold extension of US policy. Regional players understand that these controls are tightening. Many Western writers focus on the financial loss for US chipmakers. In contrast, East Asian observers focus on national AI competitiveness. For Seoul, this policy means a new focus on its own domestic AI supply chains. South Korea must build its own compute infrastructure as the global market splits.
Many Western experts thought that restricting chip shipments would stop China’s AI progress. This new rule challenges that idea. It directly targets how Chinese firms find other paths to compute. However, enforcing this rule will be hard. It is difficult to track and police cloud usage across complex global networks. This task is much harder than stopping physical shipments at a border.
To judge this policy, we must watch the investments of Chinese chip firms like Huawei and Alibaba. We should also track the capacity plans of major Chinese cloud providers. We need to monitor their new compute capacity and the specific chips they use. If these US rules work, China will invest more in its own chip factories. We will see a push for new compute architectures within the next 12 to 18 months.
This story appeared in AsiaAI.FYI Issue #80.