The US let Nvidia sell chips to an empty office in China

Introduction

Recent reporting has highlighted a troubling lapse in U.S. export controls: Nvidia’s high‑performance H200 AI chips have been cleared for sale to a Chinese company that, on the ground, looks more like an empty office than an operating business. The episode underscores how quickly policy relaxations, lobbying pressure, and opaque vetting can combine to create security gaps in the world’s most sensitive technology supply chain.

How a Hollow Office Got Approved

The Commerce Department’s licensing office granted a clearance to a firm called New Matrix, a company that was incorporated just over a year ago and whose listed address is an industrial park with no visible signage, dangling wires, and no staff. Local property managers and a building directory confirm that the firm does not appear in any tenant list. Online searches turn up only sparse references, suggesting the entity exists largely on paper.

Ties to Blacklisted Interests

What makes the clearance especially concerning is New Matrix’s ownership structure. The company is linked to a Chinese private‑equity fund that has been on the U.S. Entity List since 2024, a designation reserved for parties deemed a national‑security risk. The fund’s inclusion means any U.S. technology transferred to its affiliates could be used to bolster capabilities that the United States seeks to restrict.

Policy Shifts and Corporate Lobbying

During the Biden administration, Nvidia was barred from selling its most advanced AI chips to China. The subsequent Trump‑era policy change relaxed those limits in exchange for a 15 percent royalty on all sales, a deal that Nvidia aggressively lobbied for. The company’s CEO, Jensen Huang, has publicly championed the need for a more open market, while also navigating a complex geopolitical environment that includes alleged espionage attempts targeting him.

A Pattern, Not an Isolated Slip

The New Matrix case is not the first instance of a questionable Chinese recipient slipping through the licensing net. Past investigations have uncovered similar approvals for firms with opaque ownership or connections to sanctioned entities. Each occurrence erodes confidence that the export‑control system can reliably differentiate legitimate commercial partners from those that could divert technology to military or surveillance applications.

What This Means for Future Oversight

• Stricter Due Diligence: Agencies must verify not just corporate paperwork but also physical presence and operational legitimacy before granting licenses.
• Transparent Reporting: Companies receiving export approvals should be required to disclose basic information about their end‑users, enabling public and congressional scrutiny.
• Re‑evaluating Royalty Deals: The financial incentives that motivated policy relaxations need to be balanced against the risk of technology leakage.
• Enhanced Inter‑Agency Coordination: Commerce, State, and Defense departments should share intelligence on entities with known security concerns to prevent inadvertent approvals.

Conclusion

The clearance of Nvidia’s AI chips to a seemingly empty office illustrates how policy concessions, corporate lobbying, and insufficient vetting can converge to create security blind spots. Strengthening the export‑control process—through more rigorous on‑the‑ground verification, greater transparency, and tighter inter‑agency cooperation—will be essential to safeguard cutting‑edge technology from falling into the hands of actors that could use it against U.S. interests.