The Chip Ban Is Economic Warfare
The U.S. semiconductor export controls against China are not merely trade policy but a deliberate attempt to deny a geopolitical rival the computing substrate needed for AI supremacy. DeepSeek's success has exposed both the power and the limits of this strategy.
"What concerns me is the mindset undergirding something like the chip ban: instead of competing through innovation in the future the U.S. is competing through the denial of innovation in the past." Ben Thompson
The chip ban rests on a structural asymmetry in the semiconductor supply chain. More than 90% of chips at leading-edge nodes are manufactured in Taiwan and South Korea. All EUV lithography comes from ASML in the Netherlands. Japanese firms hold monopolies in photoresist coaters and developers. American firms dominate deposition, etch, and metrology equipment. By restricting access to these chokepoints, the U.S. can slow China's ability to manufacture advanced chips domestically. Taiwan and China have made little independent progress in semiconductor capital equipment, which means even captured fabs would grind to a halt without Western tool support and chemical supply.
But DeepSeek demonstrated the strategy's critical weakness: denied access to top-tier hardware, Chinese labs invested heavily in algorithmic efficiency and optimization. DeepSeek's V3 and R1 models achieved competitive performance with far less compute, using legal chips with clever engineering. The chip ban inadvertently created a forcing function for efficiency innovation. As Ben Thompson argues, this approach trades future innovation for present advantage and may ultimately sow the seeds for a competing chip ecosystem that undermines the very supply chain dominance the ban depends on.
The broader geopolitical picture is one of semiconductor regionalization. Governments from the U.S. to the EU to Japan to South Korea are racing to build domestic fab capacity, driving smaller, less efficient fabs that increase capital intensity across the industry. The era of globally optimized semiconductor supply chains is ending.
Takeaway: Denial strategies buy time but not permanent advantage and the efficiency innovations they force on adversaries may ultimately prove more disruptive than the denial itself.
See also: Dennard Scaling Ended and Everything Changed | Custom Silicon Will Eat General Purpose Computing | Efficiency Is The Enemy of Resilience