Getting What You Wished For: What Sun Tzu Would Say About China's Self-Imposed Chip Ban
All views in this newsletter are my own and do not represent the views of The R Street Institute, the US Navy, or any other organization I am affiliated with.
In my first newsletter, I argued that the U.S. approach to semiconductor export controls represent strategic error. While we were attacking Chinese capabilities directly, we were also strengthening their drive for technological sovereignty. My opinion was contrary to national security hawks, who demanded we get tougher, deny China more chips, and stop Nvidia from selling anything to Chinese companies.
This September, it seems the hawks got exactly what they wanted, but from a different source. China's Cyberspace Administration banned its biggest tech companies (ByteDance, Alibaba, and others) from purchasing Nvidia's AI chips. Beijing directed these firms to halt testing of the new RTX Pro 6000D and cancel pending orders, despite several companies having indicated interest in purchasing tens of thousands of these GPUs.
The hawks who demanded we deny China chips should be celebrating. Instead, China's self-imposed ban reveals exactly why their strategy was wrong all along.
What the Hawks Wanted
For years, critics attacked Nvidia for designing chips specifically for the Chinese market. After US export controls blocked sales of advanced GPUs, Nvidia created the H20 and RTX Pro 6000D. These chips complied with export restrictions by offering reduced performance compared to top-tier products. Hawks viewed even these neutered chips as unacceptable strategic concessions.
The argument went like this: any chips reaching China strengthens their AI capabilities. We need to deny them everything, force technological isolation, and watch their AI ambitions collapse without American hardware. This was framed as getting tough on China, demonstrating American resolve, protecting national security through capability denial.
In Sun Tzu's hierarchy of strategic approaches, this represents level 4 thinking where the general is attacking the enemy's army, their direct capabilities. It treats technological competition like a chess game where denying your opponent pieces ensures victory. Remove China's access to advanced chips, and you win the AI race by default. The only problem is, that's not how technological competition works in the 21st Century. Technology has been democratized to a level where import substitution can be a viable path. China's recent actions prove this explicitly.
Further, the most common objection, that these chips will end up in Chinese military systems, applies to nearly all advanced technology trade. Machine tools, advanced materials, precision equipment all have dual-use applications. Yet we monitor rather than ban these exports, accepting that there is a dual-use nature inherent to modern technology. The question isn't whether some Nvidia chips might support military applications, it's whether wholesale bans advance American strategic interests better than establishing and maintaining dependencies that create leverage. This question has been answered by China's self-imposed ban of Nvidia chips.
What We Actually Lost
When China implemented the capability denial strategy that hawks advocated for, here's what America lost:
Strategic Leverage: When Chinese companies bought Nvidia chips, they were dependent on American technology. Companies had previously maintained hope for renewed access to Nvidia products if geopolitical tensions eased. That calculus has fundamentally shifted toward building domestic alternatives. Dependencies create leverage. We voluntarily gave ours up.
Visibility: Nvidia sales provided insight into Chinese AI development. What they buy reveals what they're building. China's intensified customs inspections on Nvidia imports show that Beijing is serious about enforcing separation. We're now blind to capabilities we could previously track through commercial channels.
Revenue Funding R&D: China accounted for a substantial portion of Nvidia's global sales representing billions in annual revenue that funded American AI leadership. That money now goes to Huawei and Cambricon instead, funding our competitors.
Integration Costs: Chinese companies embedded in the CUDA ecosystem faced massive switching costs. Every AI model trained on Nvidia infrastructure, every developer skilled in CUDA, every application built on Nvidia tools represented strategic friction against Chinese independence. We removed the very friction that protected American strategic interests.
Sun Tzu's Lesson
China's ban vindicates Newsletter #1's argument: a better strategy would have been attacking their plans (Sun Tzu's level 2), not attacking capabilities (level 4).
Export controls attack capabilities—the chips themselves—rather than China's strategic objective: technological sovereignty. Every restriction validates Beijing's argument that independence is necessary. We attack supporting pillars while strengthening the center of gravity.
The right approach makes technological sovereignty economically counterproductive: Nvidia chips so integrated that switching costs are prohibitive, dependencies so deep that sovereignty means falling behind, developer ecosystems so valuable that isolation weakens competitiveness.
America didn't win the Cold War by denying Soviets all technology. We won by making Western integration more valuable than Eastern Bloc autarky. We made their drive for autarky into a strategic liability.
We're doing the opposite now. We're validating Beijing's sovereignty argument, forcing the switch we wanted to prevent, funding Chinese competitors, losing visibility, and eliminating switching costs that created strategic friction.
The Strategic Irony
Beijing asserts that Huawei and Cambricon now produce chips with comparable performance to Nvidia's China-exclusive offerings. Even if true—and market behavior suggests otherwise, since companies were ordering tens of thousands of Nvidia chips before the ban—this represents China achieving exactly what our export controls aimed to prevent.
Analysis suggests Beijing views the ban as strategic positioning, a deliberate step toward technological independence from American suppliers. One executive's assessment to the Financial Times captures the shift: the message is now unambiguous, previous hopes for renewed supply access have been replaced by comprehensive commitment to building independent, Chinese systems.
That's exactly the outcome discussed in my first newsletter. Export controls attacking capabilities while strengthening the drive for technological sovereignty.
Sun Tzu wrote: "The supreme art of war is to subdue the enemy without fighting." Making Chinese companies want to buy Nvidia chips, ensuring that desire creates strategic dependencies beneficial to America, is infinitely superior to forcing them not to buy chips at all. The first approach creates leverage. The second eliminates it.
The hawks who criticized Nvidia sales wanted capability denial. China is voluntarily giving them exactly that. This reveals why attacking capabilities rather than strategy was always the wrong approach.
What This Means For...
Policymakers: The export control hawks got what they wanted—China effectively banned from Nvidia chips. This reveals why capability denial strategies fail. We forced technological sovereignty while surrendering leverage, revenue, and visibility. Sun Tzu's lesson: attack strategy, not capabilities. Before implementing restrictions, ask: does this attack their plans or just their capabilities? What happens when they adapt? The dual-use argument, applied consistently, justifies comprehensive autarky. Applied selectively, it reveals policy driven by political visibility rather than strategic coherence.
U.S. strategic competition: We're watching what happens when you follow level 4 strategy (attacking capabilities) instead of level 2 (attacking plans). China is building an independent AI ecosystem with money that would have funded American R&D. We voluntarily eliminated switching costs and dependencies that created strategic friction. Every dollar that would have gone to Nvidia now funds Huawei. Every dependency that created leverage is gone. The hawks promised security through denial. We got vulnerability through isolation—ours, not theirs.
Tech companies: The Nvidia case study shows why capability denial strategies hurt American competitiveness. Revenue that funded R&D leadership now funds competitors. Dependencies that created leverage are eliminated. Market access isn't just profit—it's strategic positioning. When hawks demand you stop selling to adversaries, they're demanding you surrender strategic advantages.
Aspiring strategic thinkers: China's self-imposed Nvidia ban is a perfect case study in strategic hierarchy. Level 4 thinking (attacking capabilities) often strengthens what you're trying to weaken. American hawks and Chinese nationalists both made the same error: confusing denial with strategy. Sun Tzu would recognize both positions as attacking the wrong level. Real strategic thinking asks: what happens after we implement this policy? China's ban provides the answer: we get exactly what we wanted, and it proves we wanted the wrong thing.