The United States has moved to seal a significant gap in its controls on advanced artificial intelligence semiconductors, a step that could reshape how global companies obtain high-performance computing power and further test technological ties with China.
At the same time, Nvidia (NASDAQ: NVDA), the dominant supplier of those chips, has adjusted its own equity portfolio by increasing its stake in SpaceX while substantially reducing its holding in Intel (NASDAQ: INTC), signaling a broader reorientation of priorities amid evolving market and regulatory pressures.
On November 16, 2026, the Commerce Department’s Bureau of Industry and Security issued an interim final rule designed to prevent entities in restricted destinations from circumventing longstanding export limits by remotely accessing American-made chips housed in data centers located in third countries.
Previous restrictions focused primarily on the physical shipment of advanced semiconductors.
As cloud computing matured, firms headquartered in China and other destinations subject to arms embargoes were able to rent large amounts of computing capacity from facilities in places such as Singapore, Malaysia, Indonesia, Thailand, and Brazil without triggering the same controls.
The new measure requires a license when an exporter knows that covered chips or servers will be used in a data center providing more than 12 million in aggregate total processing performance to or for companies whose headquarters or ultimate parent is located in Country Group D:5 destinations (which include China, Russia, and Iran) or Macau.
Existing arrangements that meet this threshold have a 90-day window to obtain authorization once the rule becomes effective.
In parallel, the rule imposes new licensing requirements for the export, re-export, or in-country transfer of advanced chips to entities in Indonesia, Malaysia, Singapore, Thailand, and Brazil—jurisdictions the administration identified as presenting an elevated risk of diversion.
Exceptions exist for recipients headquartered in the United States or controlled by a U.S. parent, subject to reporting obligations.
Licensees must conduct robust due diligence on customers, including ownership structures, and submit periodic reports detailing parties involved and final destinations.
The measure also formally rescinds the prior administration’s broader AI diffusion framework while tightening certain rules applicable to foundries and packaging operations.
The Commerce Department described the action as necessary to ensure that existing limits on sensitive American technology cannot be sidestepped through remote computing arrangements.
These changes arrive as President Trump prepares for discussions with Chinese leadership and follow heightened scrutiny of diversion pathways, including scrutiny of certain cloud operators that had facilitated remote access.
Semiconductor markets and cloud-service providers face potential disruption: demand patterns for high-end accelerators could shift, licensing timelines may slow deployments, and operators in the five designated countries will confront additional compliance burdens.
The rule tests the durability of U.S.-China technological interdependence, as Beijing has pursued domestic alternatives while still relying on imported computing capacity for training large models.
Industry observers note that the threshold of 12 million total processing performance roughly corresponds to the capability of several hundred high-end accelerators, concentrating the new controls on substantial commercial deployments rather than small-scale use.
Against this backdrop of tighter controls on chip access and remote computing, Nvidia’s latest regulatory disclosure reveals a strategic portfolio shift that underscores the company’s expanding influence beyond traditional semiconductor customers.
In its Form 13F filed with the Securities and Exchange Commission on November 16, 2026, covering holdings as of September 30, Nvidia reported increasing its stake in SpaceX to 130.5 million shares, up from 122.8 million shares three months earlier.
Concurrently, it reduced its Intel position by 25 percent, from 214.8 million shares to 161.1 million shares.
At quarter-end valuations, the SpaceX holding stood at approximately $19.7 billion and the remaining Intel stake at roughly $19.4 billion, making SpaceX the largest single disclosed equity investment.
Together the two positions accounted for more than 80 percent of Nvidia’s reported $48.3 billion portfolio of U.S.-listed securities, down from $63.4 billion at the end of the prior quarter.
Nvidia also trimmed smaller positions in Nokia and Synopsys while leaving holdings in CoreWeave, Coherent, Nebius, and Generate Biomedicines unchanged.
The SpaceX stake originated indirectly: Nvidia invested in the artificial-intelligence firm xAI in early 2026; SpaceX subsequently acquired xAI, converting the exposure into equity in the aerospace company, which later listed publicly.
The increased holding reflects growing commercial ties, as SpaceX has indicated plans to deploy substantial volumes of Nvidia processors in its own AI infrastructure.
The reduction in Intel, by contrast, follows an earlier multi-billion-dollar investment tied to collaborative processor development and may signal a reassessment of priorities as Nvidia concentrates resources on high-growth AI and aerospace-related opportunities while the broader semiconductor landscape adjusts to export constraints and competitive dynamics.
Viewed together, the regulatory tightening and Nvidia’s portfolio moves illustrate interconnected pressures on the global semiconductor ecosystem.
Export controls that limit remote access to advanced chips constrain the addressable market for leading suppliers and encourage diversification into adjacent sectors such as aerospace, where computing demand is rising.
At the same time, the concentration of Nvidia’s disclosed investments in SpaceX and a reduced but still substantial Intel position highlights both the scale of capital flowing into AI-adjacent infrastructure and the ongoing evaluation of traditional partnerships.
As the new licensing regime takes effect and markets digest the investment adjustments, semiconductor pricing, cloud capacity allocation, and the trajectory of U.S.-China technology relations will remain closely watched.
