The Enforcer Is Now the Petitioner

The enforcement architecture the United States spent five years building was designed to keep American chips out of China. In June 2026, the US president flew Nvidia's CEO to Beijing on Air Force One to ask Xi Jinping to let them in. Trump told reporters his "first request would be to open China to U.S. businesses" ^[1]. Jensen Huang, sitting beside him, said he wasn't expecting grand declarations — "It's just going to be purchase orders" ^[1]. Meanwhile, Chinese regulators were blocking licensed Nvidia H200 chips at the border despite conditional US approval, pushing domestic Huawei alternatives instead ^[2]. Commerce Secretary Lutnick confirmed Beijing had blocked the purchases ^[2]. The side that began as the enforcer is now the petitioner. The side that spent years trying to circumvent controls is now declining chips it is finally offered. This inversion is not a diplomatic gambit. It has a material foundation: China's domestic chip alternatives have matured to the point where US chips are expendable. Self-sufficiency in AI chips rose from 10% in 2021 to 41% this year, with Morgan Stanley projecting 86% by 2030 ^[3]. On May 26, China certified nine domestic AI chips under its national security framework — the first time AI chips were included as a standalone procurement category — spanning seven manufacturers including Huawei Ascend and Alibaba Zhenwu ^[4]. Alibaba has already delivered 560,000 Zhenwu chips to more than 400 customers ^[5]. Huawei has mass-produced 381 chips based on its Tau Scaling Law, which targets 1.4nm equivalence by 2031 without EUV lithography — the precise equipment US export controls have denied China since 2019 ^[6]. Huawei's rotating chairman put the cause plainly:

If the US hadn't forced our country, our companies, and our industry, we wouldn't have done something like this. — Xu Zhijun

Nvidia's own CEO has acknowledged the result. Huang told investors Nvidia has "largely conceded China's AI chip market to Huawei" ^[5]. Even when Chinese military-linked universities attempted to lease H200s via remote compute, Nvidia's position was that China has "more than enough domestic chips for all of its military applications, with millions to spare" and that it "makes no sense for the Chinese military to depend on American technology" ^[7]. The company whose chips are being sought says China doesn't need them. The enforcement regime was designed to deny China chips it wanted. China now declines chips it is offered. That is the inversion, and it runs deeper than a single product line. While the interdiction logic flipped, both governments quietly converged on the same instrument: state power steering chip procurement toward national champions. The US government now holds a 10% stake in Intel — $40 billion, making it the third-largest shareholder — following a meeting between Intel's CEO and the president ^[8]. The administration personally brokered an Apple-Intel manufacturing partnership, with Commerce Secretary Lutnick meeting repeatedly with Tim Cook to encourage the deal ^[9]. Trump announced it June 18 as a reshoring imperative ^[9]. Google ordered more than 3 million TPUs from Intel for 2028, using Intel's EMIB packaging — a contract-manufacturing win driven partly by TSMC's packaging lines being sold out through 2027 ^[10]. Intel's foundry unit now generates $5.4 billion, 40% of revenue ^[10]. This is the same tool China is using. Beijing's $295 billion AI data center network plan, led by the NDRC, mandates at least 80% of core technology — including AI chips — be sourced from domestic suppliers like Huawei ^[11]. The five-year initiative explicitly seeks to minimize dependence on foreign chipmakers ^[11]. The US spent years condemning China's state-directed semiconductor industrial policy. It is now deploying the same instrument: equity stakes in national champions, government-brokered procurement deals, and mandates that redirect supply chains toward domestic producers. The Apple-Intel deal covers only lower-end chips so far, and TSMC retains over 90% of Apple's supply — the pivot is real but partial ^[12]. The direction, though, is unmistakable. The European Union has made this a three-sided convergence, not a bilateral one. On June 3, the EU unveiled a Technological Sovereignty Package including Chips Act 2.0 and a Cloud/AI Development Act, targeting 20% of the global semiconductor market by 2030 and restricting foreign cloud providers from public tenders in sensitive sectors ^[13]. EU tech commissioner Henna Virkkunen framed it in language that could have come from Beijing or Washington: Europe wants to "avoid risky dependencies on single dominant suppliers" ^[13]. Three major economies, independently arriving at the same answer: build at home. Enforcement has not been abandoned. The US closed the overseas-subsidiary loophole on May 31, clarifying that licensing requirements apply to all China-headquartered companies regardless of where their subsidiaries sit ^[3]. Taiwan is considering stricter AI chip export bans to China, consulting with Washington on processing-power thresholds ^[14]. Congress is weighing a bipartisan bill to subject ASML to US export restrictions, after the Netherlands declined to act on US accusations that ASML illegally shipped EUV tech to China — the Dutch foreign trade minister calling existing rules "perhaps the strictest in the entire world" ^[15]. But enforcement has become the secondary track. The administration's posture is deregulatory where it matters most: Trump canceled an AI executive order that would have required developers to submit models for government review, saying "We're leading China. We're leading everybody, and I don't want to do anything that's going to get in the way of that" ^[16]. In the same period, the government sanctioned Anthropic — the company publicly advocating stricter controls — over a Defense Department supply-chain dispute, then lifted the designation after negotiations ^[17]. The pattern is consistent: race faster, not control harder. Jensen Huang has named what is at stake. He warned it would be a "horrible outcome" if AI models around the world eventually ran best on non-American hardware ^[18]. DeepSeek has already optimized its V4 model for Huawei Ascend chips ^[18]. The competition has shifted from controlling chip trade to ensuring AI stacks run on domestic hardware — which is, precisely, the framing China uses for its substitution strategy. The original question of the chip war was: who can control the other's access? That question has been replaced. The new one is: who can build faster at home? The answer is still being written. But the side that wrote the export rules is now the one knocking on the door asking to sell.