Back

US Chip Controls Extend from Overseas Computing Power to Lithography Machines: Why ASML May Be More Impacted Than NVIDIA?

NeutralShare toX (Twitter)Telegram

Geopolitical scrutiny and national quotas are overshadowing commercial efficiency, becoming the absolute foundational hard constraint for the global semiconductor industry in the post-Jensen Huang era.

Author: Ozeco, Crack The Market

Compiled by: TechFlow

TechFlow Summary: Just days after Jensen Huang stepped down as CEO of Nvidia, the U.S. Department of Commerce's Bureau of Industry and Security (BIS) issued an interim final rule on November 16, officially extending semiconductor export controls to third-country offshore AI data centers. However, according to macro research firm Crack The Market, this is merely the opening act.

The article deeply deduces the upcoming bipartisan MATCH Act being pushed by the Senate, analyzing how it elevates controls from executive discretion into binding federal law. The report sharply points out that before compute access is truly severed, stringent restrictions will first backfire on upstream semiconductor equipment giants like ASML and Lam Research, reshaping the global compute geopolitical landscape.

Introduction: Washington's Hawkish Counterattack Against China

When Jensen Huang suddenly resigned as CEO of Nvidia last Thursday, I explicitly pointed out that Washington's hardliners on China had won an interim victory and would inevitably move quickly to lock in these gains.

On November 16, the U.S. Department of Commerce officially released a major rule targeting overseas offshore compute, equating the provision of advanced chip cloud leasing services to designated entities with physical chip exports.

If I were to summarize my core assessment in one sentence: the past six years of chip export controls successfully delayed compute acquisition, but the incoming MATCH Act will severely cripple upstream semiconductor equipment manufacturers before it truly restricts adversaries.

Investment Manager Key Takeaways

In a legal sense, the new rule is extremely broad: exporting advanced compute chips to overseas data centers in five key transit countries—Singapore, Malaysia, and the UAE—now similarly requires applying to the Department of Commerce for a specific license.

The U.S. executive branch retains significant discretionary space: licenses will operate under a case-by-case review mechanism, meaning the White House can flexibly tighten or loosen enforcement based on the pace of geopolitical diplomatic negotiations.

The upcoming MATCH Act is fundamentally different: it is a statutory regulation enacted by Congress, which will irreversibly lock in restrictions on semiconductor lithography machines and manufacturing equipment in the form of binding legislation.

During the previous semiconductor downcycle, overseas lithography giants were highly dependent on counter-cyclical purchases from specific markets. For Dutch lithography giant ASML, its shipment proportion to China once soared from 11% to 39%.

Nvidia possesses strong risk absorption capacity heading into fiscal year 2027: the self-built compute demand of native U.S. tech giants remains in a supercycle, fully capable of absorbing capacity quotas diverted due to compressed offshore channels.

The secondary market currently undervalues countermeasure risks from critical raw materials: in areas such as export restrictions on rare earth elements and critical base metals, the global supply chain still harbors massive potential volatility premiums.

What Exactly Was Outlined in the November 16 Rule?

The formal title of this administrative rule is "Addressing Remote Access to Advanced Semiconductors," filling regulatory loopholes that previously only covered hardware exports.

Figure: Screenshot of the cover page of the interim final rule document published by the U.S. Department of Commerce's Bureau of Industry and Security

The core clauses of the new rule are very clear: Any enterprise exporting advanced chips or high-bandwidth networking equipment to overseas data centers that may provide large-model training services for restricted entities must obtain prior approval.

Regulatory details comprehensively block common evasion pathways: sub-sites of distributed clusters will be calculated toward total aggregate compute, rendering attempts to bypass thresholds by splitting server rooms across multiple locations completely ineffective.

Indirect proxy training models are similarly brought under control: Overseas non-restricted clients who act as agents for restricted entities to conduct large-model pre-training will also hold colocation facilities jointly liable for compliance.

Data center operators must implement strict "Know Your Customer" (KYC) verification, thoroughly piercing through the ultimate beneficial ownership structures of the entities renting compute power.

Existing overseas data centers already providing compute services to restricted entities are granted a 90-day transition buffer period and must complete compliance license approvals within this timeframe.

How Large Is Nvidia's Actual Risk Exposure?

Nvidia officially released compliance guidance at the earliest opportunity, stating it is comprehensively assessing the rule's impact and maintaining that long-term demand has not been fundamentally disrupted.

For Nvidia, the most direct regulatory exposure concentrates on the Southeast Asian market. Over the past several quarters, regional revenue represented by Singapore accounted for roughly 10% of its total revenue.

However, in its fiscal year 2027 outlook, capital expenditure guidance from North American hyperscale cloud providers like Microsoft, Meta, and Amazon continues to be revised upward, giving them ample capacity to absorb any overflowed GPU capacity resulting from blocked offshore licenses.

Therefore, for the chip design giant, the new rule more closely aligns with rising compliance friction costs in the short term, rather than an irreversible cliff-like drop in revenue.

The MATCH Act: The True Semiconductor Black Swan

What truly holds disruptive, lethal force is the MATCH Act (Making Advanced Technology Controls Happen Act), which a bipartisan congressional bloc is jointly advancing.

Figure: Distribution of revenue among major global semiconductor equipment manufacturers over recent years and changes in proportion by key regions

Different from the interim rule signed by the executive branch, the MATCH Act is statutory legislation aimed at completely stripping the presidential administration of administrative negotiation leverage on semiconductor equipment controls, permanently codifying strict thresholds.

Once signed into law, its restrictions on deep ultraviolet lithography (DUV) machines and high-end etching and thin-film deposition equipment will no longer be subject to diplomatic negotiation flexibilities.

The sectors hit first and hardest will not be downstream compute renters, but upstream semiconductor equipment leaders like ASML, Lam Research, and Applied Materials.

Prior to offsetting lost major cash flow procurement sources with sufficient domestic fab expansion subsidies, the mandatory severance of these pipelines will force the global semiconductor equipment industry to bear the brunt of dual pressures on profits and valuations.

Read the original at Techflowpost

More News