On April 30, 2026, two seemingly independent events landed on the same day — the U.S. Department of Commerce suspended equipment supply to Hua Hong Semiconductor, and the U.S. House of Representatives simultaneously moved against Airbnb and Anysphere over their selection of Chinese AI models. One front cut into the hardware supply side, the other into the choice architecture of the software ecosystem; together they pointed to a single judgment: U.S. tech containment of China is shifting from "point sanctions" to systemic blockade.
On May 4, a third front emerged — Nvidia CEO Jensen Huang personally announced that Nvidia's share of the China AI accelerator market had fallen to zero. From supply-side hardware bans, to demand-side ecosystem scrutiny, to the supplier side's own self-examination — the three lines converge on one conclusion: sanctions are not only backfiring, they are doing so irreversibly.
On May 7, a fourth front came into view — Sputnik (Rossiya Segodnya) used SenseTime as a case study to reveal a third survival path for Chinese AI companies under the sanctions environment: lowering model-access costs to expand markets outside the United States. From "hardware catch-up" to "efficiency revolution," sanctions have produced an unexpected cost race.
U.S. chip sanctions against China have completed a systemic evolution from single-point supply-side strikes to a seven-front architecture: equipment bans + procurement audits + supplier self-examination + cost competition + AI model-access tiering + allied resistance to extraterritorial jurisdiction + supply-side confirmation of market erosion. The hardware front cuts off China's access to advanced chipmaking equipment; the software/ecosystem front blocks U.S. firms from choosing Chinese AI solutions; the supplier-CEO front confirms the self-defeating "thousand enemies, eight hundred self-loss" effect of sanctions; the Chinese-firm front documents a cost revolution that sanctions have catalyzed; AI model-access tiering institutionally excludes China from the frontier of AI capability; allies have moved from passive acceptance to formal institutional resistance; and the latest front — Nvidia CEO Jensen Huang personally admitting "the market is lost" on the same day as the ASML CEO's statement — constitutes double-ended (supplier + competitor) confirmation that sanctions are not only backfiring but irreversibly reshaping the global AI-chip competitive landscape. Seven fronts point to one structural verdict: systemic tech decoupling is accelerating irreversibly, while sanctions simultaneously give rise to a more aggressive, cost-disruptive Chinese AI ecosystem.
The Hua Hong Equipment Ban — Old Method or New Turning Point?
In an interview with Sputnik (Rossiya Segodnya), political commentator Sima Pingbang offered a three-layer reading of the U.S. Commerce Department's suspension of equipment supply to Hua Hong: irreversible dependence of U.S. firms, an uninterrupted-but-reconfigured Chinese supply chain, and the maturation of China's "software-plus-hardware parallel" systemic competition.
The Irreversible Loss of U.S. Firms
As the ban landed, shares of Lam Research, KLA, and Applied Materials dropped between 3 and 5.8 percent. Sima Pingbang pointed out that although the U.S.–China tech war has run for eight years, the semiconductor supply chain remains deeply interwoven — and losses incurred by U.S. companies cannot be made up in other markets around the world.
The China-revenue share of these three companies — especially their dependence on Hua Hong and Huawei — cannot be transferred by a stroke of an administrative pen. Equipment suppliers from Europe, Japan, and South Korea will move in to fill the gap and gain stronger competitive positions in the world's largest chip market.
The "Cannot Be Interrupted" Logic of China's Supply Chain
Sima Pingbang's core judgment is that the upstream supply chain supporting China's advanced-chip manufacturing will not be interrupted by this measure, only reconfigured. He cited two cases to support this:
- The SMIC precedent: After years of U.S. pressure, SMIC's advanced-manufacturing capability has only grown stronger — sanctions did not "kill" SMIC, but forced out greater technological resilience.
- Hua Hong's technical progress: Hua Hong has already developed advanced manufacturing technologies usable for AI chips, viewed by the market as a key milestone in China's technological self-reliance.
Sima Pingbang draws a parallel between the Strait of Hormuz — which controls more than 20 percent of global fossil-energy flows, allowing Iran to remain assertive in conflict and influence global energy markets — and U.S. chip-equipment restrictions. The latter's influence on global markets is far smaller. In other words, the U.S. Commerce Department's "old playbook" is being overestimated.
The Verification Signal of "Software-and-Hardware in Parallel"
Sima Pingbang points to DeepSeek-V4 adapting to Huawei's Ascend platform as key evidence of China's "software-plus-hardware parallel" systemic competition. When U.S. compute cards are constrained at the hardware layer, China's software-side adaptation and optimization become the core pathway around the blockade. U.S.–China competition has shifted from "who has better chips" to "who can build a better system under constraint."
Congressional Scrutiny of Firms Choosing Chinese AI — From Cutting Supply to Cutting Demand
The House Select Committee on the Chinese Communist Party joined with the Homeland Security Committee to send letters to Airbnb and Anysphere, asking both firms to explain why they chose Chinese AI models rather than American alternatives. This marks an extension of U.S. tech containment from the supply side to the demand side.
The Airbnb Story: Speed and Cost as Drivers
Airbnb CEO Brian Chesky has openly stated that, when building the company's customer-service agent, the firm tested both Chinese models and those of OpenAI and Google, and ultimately chose the Chinese AI model for its "faster speed and lower cost." In production, they also use OpenAI's models, but prioritize the more cost-effective Chinese option.
The choice itself reveals a structural reality: Chinese AI services now compete with American counterparts on utility (speed, cost). This is not an ideological selection but the market's selection.
Anysphere's Predicament: Cursor's Chinese-Model Issue
Anysphere's Cursor coding tool is widely used for its excellent code-generation, but it relies on a Chinese AI model. U.S. legislators expressed deep concern in their letters: given Cursor's popularity, "software systems used in the U.S. economy, government, and defense-industrial complex may increasingly depend on models developed by laboratories associated with, and influenced by the strategic objectives of, the Chinese government."
The letters' core logic is that America's "information-security perimeter" should not extend to private-sector technology choices — but this precisely reflects Washington's deeper anxiety: when Chinese AI models persistently outperform American options on price-performance, the market will tilt toward China on its own, and this kind of "soft dependence" is harder to manage than a hardware ban.
The Deeper Logic of the Dual-Track Contest
Read together, the two messages show not two independent events but the left and right hands of a single strategy:
| Layer | Means | Target |
|---|---|---|
| Hardware layer | Hua Hong equipment ban | Cut off China's path to advanced chipmaking equipment |
| Software / ecosystem layer | Congress probes firms' AI choices | Block Chinese AI models from permeating the U.S. economic and defense systems via commercial channels |
The first aims to choke off China's capacity for technological upgrading at the source; the second aims to stop Chinese AI models from entering U.S. economic and defense systems through commercial channels — even though that "permeation" is nothing more than rational decisions American firms make in their own market choices.
The Mirror Image of Blockade — When the "Old Method" Meets the "New Structure"
Sima Pingbang's deeper judgment is worth dwelling on: the reason this U.S. "old method" is "old" is not because the method itself is outdated, but because the China it faces has changed.
China's breakthroughs in artificial intelligence are now running on a parallel track with the United States on both software and hardware. Going forward China should add countermeasures aimed at U.S. AI technology — even institutionalize them. The day the two countries decouple their frontier technologies is bound to come.
This mirrors the blockade logic revealed in Wheels of an Iron Curtain — The Policy Logic of the U.S. Blockade of China's Auto Industry: just as auto policy combines tariffs and legislation, the chip domain deploys "equipment bans + procurement audits + cost competition" — different industries, the same systemic-decoupling strategy.
It also echoes the logic of The Indispensable Hub — Global Chemical and Green-Tech Dependence on China Amid the Hormuz Crisis: the escalation of sanctions itself accelerates the formation of substitute supply chains, and China's structural manufacturing capacity and technical-adaptation capacity make it the core supplier of that substitution.
The Policy Arc — Decoupling on Cruise Control
Placing the April 30 two-front action on a longer time axis, the trajectory of U.S. chip sanctions against China comes into clear view:
| Phase | Landmark | Blockade Strategy |
|---|---|---|
| Phase 1 (2022–2023) | October 7 chip-export-control rule | Point strikes: restrict advanced chip exports to China |
| Phase 2 (2024–2025) | AI Diffusion Rule + equipment-control expansion | Line blockade: cut equipment + materials + EDA tools |
| Phase 3 (2026–) | Hua Hong ban + procurement audits | Systemic decoupling: hardware supply + software demand, two-way blockade |
From the first systematic revision of chip export-control rules in October 2022 to the simultaneous equipment ban and procurement audit of April 2026, three years of policy evolution show that U.S. tech containment against China has entered a state of "cruise control" — no longer requiring fresh event triggers, but continuing to tighten according to an institutional rhythm preset in advance.
This means that regardless of how U.S.–China relations warm or cool in other domains, systemic decoupling in technology will keep advancing under its own momentum.
Jensen Huang's Death Notice — Nvidia's China AI-Accelerator Share Falls to Zero
On May 4 at 14:54, RT (Russia Today) reported Nvidia CEO Jensen Huang's latest statement — Nvidia's share of the China AI accelerator market has fallen to zero. Considering that as recently as two years ago Nvidia held the overwhelming majority of the Chinese AI-accelerator market, this shift amounts to a cliff-edge drop.
A CEO-Level Assessment of Sanctions' Effect
In an interview with the Special Competitive Studies Project (SCSP), a U.S. bipartisan initiative, Huang made a series of systemic judgments:
Nvidia's direct sales of AI accelerators to China have gone to zero ("We've gone to zero."). Walking away from a market like China "may not be strategically rational." Export-control policies have been "largely counterproductive." Even without U.S. AI GPUs, China remains a strong competitor in frontier AI models.
These statements come from a U.S. corporate CEO himself, not from Chinese media or external analysts. This is a first-hand effect assessment from the top of the supplier side — the "thousand enemies, eight hundred self-loss" effect of sanctions, confirmed by the executive of the affected company.
From "Hardware Decoupling" to "Full-Stack Ecosystem Competition" — Three Paradoxes in the Sanctions' Effect
Huang's statements reveal three structural paradoxes:
Nvidia's direct sales in China have fallen to zero, but the total demand of China's AI market has not disappeared. Chinese developers are increasingly relying on domestic hardware (such as Huawei's Ascend line), and the software layer — the CUDA moat — is still the final American AI-tech barrier in China not yet fully broken. Huang, however, explicitly states that "the CUDA moat will not exist forever." He acknowledges: "They [China] have cheaper energy and excellent talent. China has a large pool of science and math experts, so the number of AI researchers is also considerable — that can be considered one of their national-level advantages."
Two years ago Nvidia held the overwhelming majority of China's AI-accelerator market. After sanctions took effect, Chinese firms completed substitution at astonishing speed — not a gradual replacement, but a cliff-edge transfer. The same speed, over the next two to three years, may extend from the "hardware-substitution layer" to the "software-ecosystem layer."
Huang's policy recommendation — "policy must be dynamic, in sync with the times" — represents a "vested-interest" viewpoint: Nvidia has already lost the market, so it wants policy adjusted; but for firms that have yet to enter the Chinese market, the window for policy adjustment may already have closed. Sanctions are not only counterproductive — they are irreversible. Even if policy loosens, the Chinese market has already completed its architecture shift at the hardware layer.
Integrating with the Earlier Frame — Four Fronts Converge
This new evidence adds a third front to the "decoupling framework" opened in this page — the supplier-side CEO's self-reflection front. The fourth front that followed on May 7 — SenseTime's cost revolution — completes the evidence chain of efficiency revolution catalyzed by sanctions.
| Front | Dimension | Core Evidence | Sanctions' Effect |
|---|---|---|---|
| Front 1 | Hardware supply chain (Hua Hong) | Sima Pingbang deep analysis | Sanctions forced substitution |
| Front 2 | Software ecosystem (Airbnb / Anysphere) | House committee inquiry letters | Market demand tilts to Chinese models |
| Front 3 | Supplier-side feedback (Nvidia) | CEO confirms zero market share | Sanctions' self-defeating declaration |
| Front 4 | Cost competition (SenseTime) | Sputnik report | Sanctions catalyze efficiency revolution |
The four fronts point to one conclusion: U.S. chip sanctions have accelerated China's substitution on the technology side, inflicted permanent losses on U.S. firms on the market side, and catalyzed a more aggressive Chinese AI ecosystem on the cost side — the last of these confirmed by the financials and strategic-pivot direction of Chinese industry.
The Cost Revolution — SenseTime's "Unified Multimodal" Survival Strategy
On May 7 at 19:40, Sputnik (Rossiya Segodnya) published a long analysis using SenseTime as a case to reveal a third response path for Chinese AI firms under sanctions — lowering model-access costs to expand markets outside the United States.
SenseTime's Pivot — From Selling Technology to Selling Cost
SenseTime, having been placed on the U.S. Treasury's restricted-investment list, has publicly denied the allegations while pursuing a strategy diametrically opposed to Nvidia's: rather than trying to break through the hardware blockade, it pushes model costs to the floor on top of existing domestic hardware.
Chinese company SenseTime, hit by U.S. sanctions, has publicly stated that its priority is to lower the cost of multimodal models, so as to remain competitive and expand markets outside the United States.
In late April 2026, SenseTime announced the release and open-sourcing of its SenseNova U1 family, positioned as a "unified" multimodal solution — handling language and image processing within a single architecture. This approach improves performance and lowers operating costs. The economics were already visible in the 2025 financial results: net loss narrowed in the second half of the year, with EBITDA turning positive.
The "Performance-Adequate" Strategy — Three Layers of Cost Competition
SenseTime's case exposes the logic of the fourth front, distinct from the first three (supply-side blockade, demand-side audit, supplier-side loss): SenseTime represents adaptive evolution on the survival side:
SenseTime's model-optimization direction is not to match Nvidia's H100 or B200, but to deliver "good enough" performance on existing domestic chips. The "performance-adequate" rather than "performance-leading" strategy signals that Chinese firms have accepted the reality of not surpassing the U.S. on hardware in the short term, and are instead bridging the gap through software optimization.
In a market backdrop of model saturation and intense price competition, low model-access cost becomes a competitive weapon in its own right. SenseTime's logic: while U.S. firms keep chasing parameter scale and compute ceilings, Chinese firms shift to competing on user-acquisition cost — whoever lets more users access AI services at lower prices wins greater market coverage and a stronger data flywheel.
SenseNova U1 handles both language and image processing within a single architecture, reducing the hardware needs of multi-model deployment. This is architectural innovation forced by hardware constraints — when compute cost cannot be scaled indefinitely, efficiency optimization becomes the only way out.
Structural Summary of the Four Fronts
| Front | Dimension | Core Evidence | Signal Direction |
|---|---|---|---|
| Front 1 | Hardware supply chain | Hua Hong equipment ban + Sima Pingbang's analysis | Sanctions force substitution |
| Front 2 | Software ecosystem | Airbnb / Anysphere House inquiry | Market-driven ecosystem choice |
| Front 3 | Supplier-side feedback | Huang's "zero share" declaration | Sanctions backfiring |
| Front 4 | Cost competition | SenseTime SenseNova U1 + cost-led pivot | Sanctions catalyze efficiency revolution |
The fourth front is special: it is not a passive reaction to sanctions, but an active strategic pivot — U.S. sanctions have pushed Chinese firms from a "chaser mindset" to an "efficiency-first mindset". While U.S. firms keep adding chips to the compute arms race, Chinese firms are building a survival philosophy of "doing more with less."
Huang said the sanctions are "counterproductive," and SenseTime's financials validate the point: the sanctions have not strangled China's AI industry but produced two unintended effects:
Direct effect — Inability to access the latest hardware → forced to push software optimization and architectural innovation to the extreme → produced model capabilities such as "unified multimodal," "cost competition," and "performance adequate" that apply to broader markets;
Indirect effect — U.S. high-end GPUs cannot reach China → Chinese firms pivot to serving Southeast Asia, the Middle East, Africa, and other emerging markets → Chinese AI models gain penetration in these regions that Nvidia's high-end chips cannot reach.
Sputnik's closing line — "all of this has contributed to the formation of China's autonomous AI ecosystem: although access to U.S. chips has narrowed, Chinese firms can compensate through model optimization, architectural solutions, and reliance on large platforms that can rapidly turn technology into mass-produced products" — forms a double-ended confirmation with Huang's judgment (front 3) from supply-side and demand-side perspectives.
The European Echo — Tech Decoupling Spreads Institutionally from the U.S. to the EU
In early May, a similarly "security-risk"-based restriction landed in Europe — the European Commission decided to ban EU funding support for projects using Chinese inverters. If earlier chip decoupling was a U.S.-led "point strike," then the EU's restriction on Chinese photovoltaic inverters marks tech decoupling spreading institutionally from the U.S. to the EU.
What the Decision Says
The European Commission has moved to bar EU funding from going to projects using Chinese-made inverters, citing concerns that such projects could pose security threats to the European grid — even triggering large-scale blackouts. The decision, confirmed on May 4, reflects Brussels' intensifying amplification of a so-called "concern": Europe's dependence on Chinese green technology is making it increasingly vulnerable to "security threats."
The restriction applies to EU funds directly managed by the European Commission, as well as funds managed by institutions such as the European Bank for Reconstruction and Development; it does not apply to procurement directly conducted by EU member states. Chinese inverters already installed across Europe can continue to operate.
The Chain of Diffusion
From the U.S. chip sanctions against China to the EU's inverter restriction, the diffusion of tech decoupling displays a clear institutional contagion path:
- Reproduction of the narrative frame — the chain "security risk → remote access → data leakage → infrastructure vulnerability" is identical to the FCC's restrictions on Chinese telecom equipment (the U.S. surveillance frame) and the U.S. "backdoor" allegations against Huawei (communications infrastructure). The EU replicated this narrative template on inverters, merely swapping "surveillance backdoor" for "remote emergency cut-off switch."
- Tolerance for evidence gaps — the report notes that "even German media themselves admit that there is no known case of Chinese-made inverters being used to shut down parts of the European grid." Yet the EU still pushed the restriction forward. This "better to over-warn than to under-warn" precautionary posture is identical to the U.S. approach to chip sanctions.
- The visible economic cost — a joint report by the EU Chamber of Commerce in China and KPMG points out that if the EU's revised Cybersecurity Act passes, the EU could face economic losses of up to €367.8 billion, with Germany alone bearing 46.4 percent (€170.8 billion). This mirrors the "irreversible loss of U.S. firms" (the first layer in Sima Pingbang's analytical framework) — the same institutional paradox: those imposing the restrictions bear the self-harm cost.
"The Huawei Siege of the Electrical Domain"
The EU Cybersecurity Act Revision — Systematic Exclusion of Chinese Hardware and a Five-Year Teardown Plan
On May 8, a joint report by the China-EU Chamber of Commerce and KPMG revealed a systemic upgrade to EU tech-decoupling legislation: the EU is pushing a revision of the Cybersecurity Act that, at the framework level, prohibits Chinese firms from participating in EU critical-infrastructure construction, and requires the dismantling and replacement of large amounts of Chinese hardware within five years — at an estimated total cost exceeding €367.8 billion.
The earlier inverter-funding ban (Section 7) was an administrative-level project restriction, whereas the Cybersecurity Act revision is framework-level legislation — once passed, it will legally lock in the premise that "Chinese hardware equals security risk," providing a legal foundation for all subsequent domain-specific restrictions.
Three Layers of Difference from the Inverter Ban
| Dimension | Inverter Funding Ban | Cybersecurity Act Revision |
|---|---|---|
| Layer | Administrative directive | Framework legislation |
| Scope | Specific project funds | All critical infrastructure |
| Timeline | Future projects only | Existing hardware must also be removed and replaced (five-year window) |
The Five-Year "De-Sinicization" Schedule
The report's most critical quantitative finding: dismantle and replace large amounts of Chinese hardware within five years. The schedule means several things:
- Official confirmation of stock risk. If Chinese hardware merely posed "potential risk," an administrative directive limiting new procurement would suffice. Requiring the removal of existing hardware is logically equivalent to admitting that "Chinese hardware already constitutes actual threat" — even though no public evidence supports the judgment.
- Institutional avoidance of compensation liability. The five-year schedule allows equipment replacement to proceed as "compliance retrofitting" rather than "expropriation," legally avoiding direct compensation obligations for the Chinese vendors' installed base. This echoes the "backdoor-style replacement" that Huawei 5G equipment experienced in Europe in 2019 (ostensibly not targeting any specific supplier, but with de facto exclusionary effect).
- A re-reading of the €367.8 billion cost. Tied to the explicit five-year schedule, this means roughly €73.5 billion per year in ongoing compliance costs, to be borne directly by European governments and utilities.
Beijing's Formal Response
Spokesperson Lin Jian stated that competent Chinese authorities had already submitted assessment opinions to the EU side on the proposed Cybersecurity Act revision and other matters, expressing China's solemn concern and political position. If the EU insists on enacting the law and discriminating against Chinese firms, China will take resolute measures to safeguard the legitimate rights and interests of Chinese firms.
Beijing's response carries two notable signals:
- Earlier warning. Unlike past practice, when countermeasures were activated only after a regulation took effect, China submitted its assessment opinions during the legislative push — evidence of stronger anticipation of the EU legislative path.
- "Resolute measures" left open. China has already begun to treat the EU, alongside the United States, as a "second front of tech decoupling."
Structural Mirror with Section 7
The Cybersecurity Act revision and the inverter-funding ban were disclosed on the same day (May 8), forming a double-front verification:
- Inverter restriction: funding restriction on specific equipment (tactical layer)
- Cybersecurity Act revision: legislative lock on the overall architecture (strategic layer)
When institutional barriers at the legislative level and funding restrictions at the administrative level advance in parallel, market-access obstacles for Chinese firms in the EU will no longer be surmountable individual problems.
Chang'anjie Zhishi · 15:45 — China-EU Chamber of Commerce and KPMG report: EU plans to ban Chinese firms from participating in critical infrastructure, with a five-year teardown and replacement of Chinese hardware, total cost €367.8 billion
Photovoltaic inverters are the "brain" of a PV power system. In a market where 61 percent of Europe's inverter imports come from China and 80 percent of newly built European solar systems depend on Chinese inverters, banning the "brain" from EU funding is in essence replicating the "Huawei siege" model in the electrical-equipment domain:
Step one: Security-narrative construction ("remote backdoor" → "Europe-wide blackout")
Step two: Funding/procurement restrictions (EU funding ban)
Step three: De-Sinicization of the supply chain (push for European-domestic capacity replacement)
These three steps closely resemble the path by which Huawei-led 5G equipment was gradually squeezed out of European countries between 2019 and 2023. The only difference is that the technological complexity and global manufacturing concentration of inverters is far lower than that of 5G — European local firms claim they "can scale up capacity within a few months to meet market demand," at a price increase of only about 2 percent. Inverters may be the first case where Europe's "de-Sinicization" supply-chain experiment actually lands.
Anthropic Refuses Chinese AI Researchers' Access — Institutional Tiering of AI Model Permissions
On May 12, 2026, a fifth front of tech decoupling landed in Silicon Valley. Sputnik reported an event of institutional significance: U.S. firm Anthropic refused a request by Chinese researchers to access its latest AI model Mythos.
Event chain: Anthropic announced the Mythos model in April, noting its effectiveness in software-vulnerability discovery. The model's access rights follow a tiered system — provided only to U.S. government agencies and selected organizations, with Japan in active negotiations and the U.K. government granted access at the end of April. A Chinese analytics firm requested broader access at a Singapore conference in April; the request was rejected by the U.S. side. The U.S. National Security Council expressed strong concern over the contact, characterizing it as "China attempting to seek a solution for U.S. AI business."
| Chip-Sanction Layer | AI-Model Access Layer |
|---|---|
| Equipment bans (ASML / EUV) | Model-access tiering (government / allies / excluded) |
| Foundry restrictions (TSMC) | Training data and compute controls |
| Talent restrictions (science visas) | Research-collaboration exclusion |
This event reveals that tech decoupling is extending from the hardware layer (chipmaking equipment, EDA tools) to the software and algorithm layer (AI model weights, access permissions). Anthropic's "tiered authorization" model — government level → ally level → excluded level — is essentially institutional tiering of AI capability, with the same logic as the chip export-control "license-or-embargo" binary framework, but more granular and harder to bypass (because model-access permissions can be adjusted at any time, whereas hardware exports require physical shipping).
This new front completes the five-front sanctions architecture: from hardware supply-chain severance (Hua Hong) to software ecosystem audit (Airbnb / Anysphere), from supplier-side CEO self-examination (Nvidia's zero share) to cost-competition revolution (SenseTime), to institutional tiering of AI model-access (Anthropic) — every layer accelerates the irreversibility of tech decoupling.
① Whether Anthropic continues to expand its "approved list" beyond the U.K. and Japan; ② whether an indirect "detour route" through third parties (e.g., U.K. research institutions) exists to acquire access; ③ whether other U.S. AI firms (OpenAI, Google DeepMind) follow the same tiered-access model.
The Netherlands Formally Protests the Match Act — Allies' Escalation from Passive Acceptance to Institutional Resistance
On May 14, 2026, Guancha reported an institutional-significance development: the Dutch government has formally objected to the U.S. Congress's proposed Multilateral Hardware Technology Controls Coordination Act (Match Act), opposing the law's extraterritorial jurisdiction over ASML's China business.
What the Match Act Says
The bill, jointly proposed by bipartisan U.S. lawmakers, would prohibit ASML from exporting immersion deep-ultraviolet (immersion DUV) lithography machines to China, and from providing after-sales service to its existing customers in China. On top of the total EUV ban, it would further seal off DUV equipment exports and maintenance channels.
From the October 7, 2022 chip-export-control rule (executive order) to the 2026 Match Act (congressional legislation), sanctions instruments have risen from a flexible presidential administrative tool to a rigid legislative mechanism — once enacted, the bill would impose greater institutional barriers on exemptions and transition-period dilemmas.
The Dutch Government's Three-Layer Objection
In a written reply to U.S. legislators, Dutch Foreign Trade and Development Cooperation Minister Schreuel Scheltema raised objections on three levels:
- Sovereignty limit: Opposition to the extraterritorial effect inherent in the bill — "every country has the responsibility to formulate and implement its own export-control regulations," and the United States should not impose cross-border jurisdiction over foreign firms through domestic law.
- Economic-impact assessment: Such sweeping measures would not only hit ASML's revenue and weaken its market position but could also "undermine the predictability of the trade and investment environment."
- Diplomatic-objection level: During last month's Dutch royal visit to the United States, the Dutch Prime Minister, Foreign Minister, and trade department had already conveyed this position to the U.S. side — an escalation from technical opposition to institutional engagement at the highest diplomatic level.
The Match Act touches the most sensitive nerve of allied export-control systems — extraterritorial jurisdiction. In past chip sanctions against China, the United States mainly relied on "long-arm" tools (such as the October 7, 2022 rule's restrictions on global firms using U.S. technology), and although allies objected, they lacked formal legal channels for protest. The Match Act seeks to permanently legislate that logic, triggering allies' institutional-response mechanism through formal diplomatic channels.
The Dutch protest itself will not directly block the Match Act, but it creates an important institutional precedent — an ally openly opposing U.S. extraterritorial sanctions legislation — buying political space for exemption clauses in the next round of negotiations.
Structural Mirror with Earlier Lines
The Dutch formal objection forms a continuous narrative line with Europe's earlier reactions to tech decoupling:
| Date | Event | Nature |
|---|---|---|
| May 8 | EU bans funding for Chinese inverters | Unilateral restriction |
| May 10 | Japan–EU dialogue focuses on reducing dependence on China | Stated intent |
| May 14 | Netherlands formally opposes Match Act extraterritoriality | Institutional resistance |
These three events are spread across two weeks but trace a clear progression from "passive acceptance" to "active discussion" to "formal opposition." The forceful push of the Match Act may accelerate this progression — as U.S. extraterritorial jurisdiction moves from executive order to congressional legislation, allied tolerance thresholds face a systemic test.
Huang Admits "The Market Is Lost" — Self-Confirmation of Sanctions' Effect and Forced Acceleration (increment, 2026-05-21 15:14)
On May 20, Nvidia CEO Jensen Huang gave a rare admission in a CNBC interview about the China AI-chip market: "Our competitive edge in the China AI chip market has been substantially eroded." He added that Nvidia has told investors not to hold out any hope for future approval to sell advanced chips into China.
The Significance of Huang's Admission
Huang is not a policy analyst, not a government official — he is the CEO of the world's largest AI-chip company. His "the market is lost" judgment is not a policy review but a fact-based statement on the competitive landscape. When the top executive of this company personally says "the competitive edge has been eroded," the signal is this: the export-control policy's effect in the target market has moved beyond the stage of "creating inconvenience" into "substantially reshaping market structure" — either because Chinese firms have broken through sanctions, or because substitutes have grown large enough to absorb demand.
Unlike the sanction logic discussed in earlier sections (U.S. restriction → Chinese catch-up → gap exists but narrows), Huang's framing provides an important quantitative anchor: if even the Nvidia CEO believes he "cannot compete" in the China market, then China's substitute chip capability has gone beyond "catching up" — at least in specific market segments (AI inference, edge computing, certain training scenarios), it has reached a level at which Nvidia products no longer hold an absolute competitive edge.
The Narrative Significance of "Don't Hold Out Any Hope"
Huang's words are addressed not to the Chinese market but to Wall Street and investors. When a company explicitly tells investors "do not expect us to be able to sell advanced chips to China in the foreseeable future," it is doing three things: first, managing expectations (lowering revenue expectations for China sales); second, signaling to the government (sanctions are already effective, no need for further escalation); third, building a hedge narrative (if future sales numbers keep dropping, it is not because the company is failing, but because the policy is irreversible).
The temporal symmetry of Huang's admission and the ASML CEO's "China must be self-reliant" comment
On the same day (May 21), another same-theme report appeared — Guancha 11:23 reported ASML CEO saying in an interview that "China will accelerate its self-reliance — this is an existential question," and using the metaphor of "a desert vegetable garden": when technology sources are cut off, even barren soil must grow crops. The ASML CEO (supplier-side view) and Jensen Huang (competitor-side view) on the same morning separately admitted that their own market position is shaken and that China's self-reliance is accelerating — placed side by side, the two reports compose a panorama of the sanction's effect and side effects. If the earlier sections discussed "what sanctions are doing," this section discusses "what sanctions have already caused" — a market where even Nvidia's CEO no longer holds out hope.
RT · 15:14 — Huang: Nvidia has lost ground in China's AI-chip market (deep)
Guancha · 11:23 — ASML CEO: China will accelerate self-reliance, this is an existential question (deep)
From "Lost Ground" to "Total Victory" — The Tension in Huang's Two Narratives Within Two Days (increment, 2026-05-22 11:14)
On May 22, Jensen Huang reappeared in a mainstream-media interview — this time on Fox News. Within two days, Huang made headlines on CNBC and Fox, but the tone of the two narratives was diametrically opposite.
May 20 (CNBC): "Our competitive edge in the China AI chip market has been eroded." — an admission of market loss, managing investor expectations, signaling the sanctions' effect.
May 22 (Fox): "President Trump wants us to win on every front of AI." — projecting full-spectrum AI confidence, binding to Trump's governing narrative, signaling "America is winning."
The tension between the two narratives lies not in their truth value — both can be true in different markets simultaneously — but in their audience. CNBC's audience is Wall Street investors; Huang needs to lower earnings expectations for the China market. Fox's audience is Washington's policy circle and conservative voters; Huang needs to consolidate Nvidia's "national champion" status in America's political narrative. The same person, the same company, within forty-eight hours — telling two different stories to two different markets. This is not lying; it is multi-layered narrative management.
Huang added in the Fox interview: "The President has made it clear that he wants Nvidia to make sure American tech companies and businesses can take full advantage of our latest technology and our most comprehensive services. And that is indeed happening." The core message here is not AI but "make sure American companies fully use" — implying an accelerating market-segmentation trend: advanced AI chips become a resource exclusive to American firms; Chinese firms, even if granted licensing channels, are effectively excluded.
Two days ago, Huang was saying "don't hold out any hope"; two days later, "win on every front" — from defensive expectation management with the market to offensive confidence display with the political narrative. This switching itself may be Nvidia's survival strategy under simultaneous pressure from Washington and Wall Street: in front of Washington, play America's AI tip-of-the-spear; in front of Wall Street, manage expectations of losing the largest market.
RT · 11:14 — Huang relays Trump's high hopes for Nvidia: total victory on every front of AI (aggregation)