Geostrategic Media Political Commentary, Analysis, Security, Defense
China’s chip strategy is focused on self-reliance. The United States should reconsider its decision to allow Nvidia to export H200 chips and instead enact a total export ban.
US President Donald Trump’s recent announcement to permit Nvidia to export its high-end H200 chips to China has triggered another round of heated debate on its implications for national security as well as overall strategies of competing and winning the artificial intelligence (AI) race with China. The controversy is familiar and almost ritualistic. Supporters warn that restricting exports undermines American firms and accelerates China’s domestic substitution; critics argue that any relaxation strengthens China’s AI capabilities and erodes US technological leadership.
Yet what is striking about the debate is not its intensity, but its dated premises. The core assumption on both sides—that US export controls can meaningfully shape China’s strategic trajectory—has become increasingly detached from reality. The H200 decision matters at the margin, but the larger direction of China’s policy no longer hinges on whether Washington tightens or loosens restrictions. The debate persists, but the strategic ground beneath it has shifted.
The Irony After DeepSeek: From Condemning US Controls to Targeting Nvidia
The irony began earlier this year, not long after China’s “DeepSeek moment,” when Chinese AI developers publicly signaled growing confidence in their large-model capabilities. For years, Beijing had fiercely condemned US chip export controls, while Chinese companies and government actors actively sought Nvidia’s most advanced chips through every available channel. Yet almost simultaneously, China began moving in the opposite direction.
Beijing initiated actions that looked very much like an attempt to impose its own constraints on Nvidia, citing national security concerns and pursuing the company for alleged violations of China’s anti-monopoly law. Chinese regulators accused Nvidia of abusing market dominance and of failing to meet commitments tied to earlier acquisitions, opening the door to fines and regulatory pressure.
This apparent contradiction—condemning US restrictions while tightening scrutiny of Nvidia—reveals something fundamental. China’s concern is no longer simply access to American chips. It is the political, strategic, and security implications of dependence on them.
A Chilly Reception to H200: Emergency Consultations Instead of Celebration
China has not expressed excitement about, nor even openly welcomed, the US decision to lift the ban on H200 exports. This is notable. The H200 is Nvidia’s second-best AI chip, surpassed only by the Blackwell generation, and critics in the United States argue that allowing its sale could significantly boost China’s AI development.
Instead of public enthusiasm, China’s regulatory authorities convened a series of emergency meetings with representatives from Alibaba, ByteDance, Tencent, and other major technology firms. The companies were asked to assess their demand for H200 chips—chips for which China currently has no true domestic alternatives. Officials made clear that the government would review these assessments before announcing any decision.
This response alone suggests that Beijing does not see the H200 question as a commercial issue. It is a strategic one.
A Policymaker’s Dilemma: Capability Today vs. Sovereignty Tomorrow
These meetings underscore the central dilemma facing Chinese policymakers. On the one hand, China urgently needs advanced computing power to train frontier AI models, expand inference capacity, and sustain momentum in a race it views as decisive for national power. On the other hand, reliance on US chips creates vulnerabilities that Beijing now finds increasingly intolerable.
This dilemma is not merely technical. It is political and ideological. A state that defines technological autonomy as a pillar of national rejuvenation cannot accept a situation in which its most advanced industries depend on inputs that can be restricted, monitored, or withdrawn at another government’s discretion.
The AI Race as a Sovereignty Question
The United States and China are locked in a fierce AI race, and both sides believe the outcome will shape not only national security but the future international order. For China, the experience of repeated US export controls and technology chokepoints has had a profound effect. Beijing has drawn a clear lesson: dependence is dangerous.
As a result, “domestic substitution” (国产替代) has become the core of China’s grand strategy. Semiconductor self-reliance is no longer treated as an economic objective alone, but as a sovereign imperative. In Beijing’s strategic calculus, tech sovereignty is now almost indistinguishable from sovereignty itself.
State Procurement as Industrial Policy: Locking in Domestic Substitution
Two recent developments illustrate the depth of this commitment.
First, China’s technology ministry added AI processors from Chinese firms to its government-approved supplier list for the first time, according to the Financial Times. The policy is designed to force public-sector adoption of domestic chips and could redirect billions of dollars toward local manufacturers. Crucially, this move came before President Trump’s announcement on H200 exports.
Its significance lies not only in revenue. Government procurement in China functions as a powerful industrial policy tool. By mandating the use of domestic chips, Beijing accelerates learning curves, absorbs early inefficiencies, and signals to the entire ecosystem—local governments, universities, state-owned enterprises, and private firms—that domestic substitution is a political requirement, not a market experiment.
State Capital on an Unprecedented Scale
Second, Beijing is reportedly considering pouring as much as $70 billion into its semiconductor industry, potentially the largest state-backed chip support program in the world. Whether or not the final number reaches that level, the direction is unmistakable. China is prepared to mobilize state-scale capital to overcome structural disadvantages in fabrication, tooling, and research and development (R&D).
Unlike the United States, China cannot rely on deep and flexible capital markets to finance long-horizon, high-risk semiconductor research. State funding, therefore, becomes indispensable. It allows Beijing to sustain long development cycles, tolerate failure, and coordinate supply, demand, and infrastructure in a way that private markets cannot.
Financial markets have already reacted. Shares of domestic chip firms have surged in anticipation of policy support. Even Moore Threads—a relatively obscure graphics processing unit (GPU) startup founded in 2020—has seen its stock skyrocket in recent days.
Why the Old Export-Control Debate No Longer Holds
The argument that US export controls might spur China to innovate was plausible in 2023, when the Biden administrationfirst imposed sweeping restrictions on advanced chips and manufacturing equipment. At that time, it was reasonable to ask whether such controls might backfire by accelerating China’s technological push.
That question is now largely moot. China’s determination to achieve semiconductor self-reliance no longer depends on US policy choices. Whether Washington tightens controls, loosens them, or oscillates between the two, Beijing has decided to commit all available resources to building an independent chip ecosystem. The strategic die has been cast.
The “Addiction” Theory and Its Limits
Some policymakers and advisers—including Trump’s AI and crypto czar David Sacks—have argued that allowing broad sales of non-cutting-edge chips could keep Chinese AI developers “addicted” to US technology, preserving Nvidia’s market share and limiting the rise of competitors such as Huawei. I previously advanced a related argument, calling it a strategy of dynamic precision export control.
But this logic presumes that China is willing to accept such dependence. That assumption no longer holds. After repeated policy reversals and strategic signaling from Washington, Beijing now understands the subtle calculations behind US export policy all too well.
There is an old Chinese saying: “She who breastfeeds you is your mother.” China understands that the United States is, at most, a wet nurse—not a mother—to its AI industry. Beijing fears that Chinese AI developers could become so dependent on American “milk” that they accept the US as mother, that is, subject themselves to American influence or control. It also understands that a wet nurse never provides her best milk if doing so risks empowering another person’s child too much. China will not allow itself to be trapped in such a relationship.
Precision Import Control: Using Without Submitting
Yet American chips remain enormously attractive. The H200 is reportedly six times more powerful than the H20 previously allowed for sale to China, and Chinese firms have gone to great lengths—including smuggling and gray-market procurement—to obtain restricted Nvidia chips.
Beijing also has an immediate political objective. To catch up with or surpass the United States in AI, China needs breakthroughs—more “DeepSeek moments”—to boost morale and mobilize techno-nationalist sentiment. Advanced chips help generate those moments.
This leads to Beijing’s likely path: precision import control. China will make selective, strategic use of American chips without surrendering to US standards, rules, or long-term dependence. At the same time, it will continue preparing to wean its AI industry entirely from the United States. This is not a contradiction; it is a strategy.
The Trojan Horse Problem: Chips as a Security Risk
Even so, another dilemma looms large. A central concern is Nvidia’s alleged testing of “location verification” technology—sometimes described in China as a “GPS backdoor.” The fear is that chips could be remotely locked, disabled, or monitored if used in restricted locations such as China.
Although Nvidia insists it has no “kill switch,” the mere possibility of such functionality has raised deep national security concerns in Beijing. In Chinese discourse, these chips risk being seen as “Trojan horses”—hardware that carries hidden strategic vulnerabilities.
This concern cuts to the heart of China’s calculus. If imported chips can be monitored, constrained, or weaponized through software, then dependency is not merely economic—it is political and security-related. This fear reinforces Beijing’s resolve to accelerate domestic alternatives, even at significant cost.
China’s Fixed Strategy—and America’s Only Rational Response
In short, China’s strategy is clear and determined: precision import control, buying time, domestic substitution, self-reliance, self-sufficiency, full technological sovereignty, and ultimately catching up with or surpassing the United States. And this strategy will not vary with US export-control policies. Just as with rare earths, building supply-chain security quickly and decisively is a must for the United States, regardless of what China says or does.
If that is the case, the United States has only one rational choice: like it or not, a total export ban to China. Partial measures merely delay the inevitable while transferring capability and experience to a competitor determined to decouple anyway.
Making Total Control Work: Four Necessary Accompanying Strategies
For such a strategy to be effective, four accompanying measures are essential.
First, the US Congress must pass a total ban into law. Given the capricious and transactional tendencies of recent US leadership, statutory clarity is indispensable.
Second, the United States must sign and enforce binding agreements with all countries not subject to the ban, preventing re-exports and transfers to China.
Third, Washington must strategically support Nvidia and other US chip makers in expanding global market share. US capital markets function well, but AI bubbles may burst, and public funding is limited compared with China’s massive state-backed support.
Fourth, the United States should restrict institutional investment in China’s AI sector while adopting policies that attract foreign capital into American AI, energy, and infrastructure—helping build the American AI stack according to American standards and rules.
Beyond Chips: Toward a Broader Code War
Like it or not, these strategic choices are dictated by the nature of the AI race between the world’s two great powers. Nobody can change that nature for the foreseeable future.
I believe this competition is best understood as a code war—originating in technology but extending far beyond it. It is fought not only between states, but across global markets, institutions, and domestic societies. It involves values embedded and codified in AI models and ecosystems, the codification of AI standards and rules for humanity, and the codification of social norms that ultimately determine the kind of country the United States will become in the era of AI. Elaborating on that code war lies beyond the scope of this article.
About the Author: Jianli Yang
Dr. Jianli Yang is a Research Fellow at the Kennedy School of Government at Harvard University. He is the Founder and President of Citizen Power Initiatives for China and author of For Us, The Living: A Journey to Shine the Light on Truth and It’s Time for a Values-Based “Economic NATO.”
