SI Agent News
ConfirmedChipsPolicySecurity

US approves Nvidia chip sales to a shell company in China

A Chinese firm with ties to a blacklisted entity was cleared to buy Nvidia H200 chips despite having no physical presence.

15-percentNvidia's sales cut to the US government

The short version

Why it matters

This situation suggests that current export controls may have significant loopholes that allow restricted entities to acquire high-end technology. It raises questions about the effectiveness of government vetting processes for sensitive AI hardware.

A hollow shell for AI chips

The US government has authorized the sale of Nvidia H200 AI chips to a Chinese firm called New Matrix. However, reports suggest the company may not actually exist in a functional capacity. Investigations into the firm's headquarters revealed a building that appears to be a "hollow gray shell" with dangling wires and no lights.

When reporters checked the building's directory, New Matrix was nowhere to be found. Furthermore, the property management agency for the industrial park stated that they have no record of the company at that location. This lack of physical presence raises questions about how the firm qualified for government approval to handle high-end hardware.

Despite these red flags, the company remains on the list of approved entities. The situation suggests a breakdown in the due diligence process intended to prevent sensitive technology from reaching restricted parties. It highlights a discrepancy between official government approvals and the reality on the ground in China.

Links to blacklisted entities

The scrutiny surrounding New Matrix is intensified by its ownership structure. The company's owners are associated with a Chinese private equity group that the US government placed on a restricted list in 2024. This means the firm is officially recognized as a risk to national security or trade interests.

The fact that a company linked to a blacklisted entity received clearance to purchase Nvidia chips is a significant oversight. It suggests that the vetting process may be failing to catch connections to prohibited organizations. Such lapses could allow restricted actors to acquire the computing power needed for advanced AI development.

New Matrix was created only 17 months ago, making it a relatively new player in the market. The lack of online information and the absence of a physical office make it difficult to verify its business purpose. These factors combined point toward the company serving as a front for other interests.

Shifting trade policies

The current landscape for Nvidia's sales to China was shaped by a shift in administration policy. While the Biden presidency placed strict restrictions on chip sales, the Trump administration relaxed those limits. This change occurred after Nvidia agreed to provide the government with a 15-percent cut of all sales to the region.

This agreement marked a pivot in how the US manages the export of high-end semiconductors. By allowing more sales in exchange for a revenue share, the government opened a door that is now being exploited. The New Matrix situation serves as a specific example of how these broader policy changes can be navigated by shell companies.

Nvidia's position in this trade dynamic is complex. The company has engaged in intense lobbying to maintain its market access while navigating various security concerns. These efforts have occurred alongside reports of the company being the subject of a Chinese state security spying operation.

A pattern of oversight lapses

The issue with New Matrix is not an isolated incident according to recent reporting. There appears to be a recurring trend of businesses with ties to organizations facing national security trade bans being granted permission to buy Nvidia chips. This suggests a systemic weakness in the Commerce Department's vetting procedures.

When multiple firms with ties to restricted entities receive clearance, the effectiveness of export controls is called into question. It indicates that the mechanisms designed to protect sensitive technology may be easily bypassed. These gaps allow for the potential flow of advanced hardware to actors that the US government has explicitly tried to block.

The Commerce Department is now facing pressure to improve its due diligence. The goal is to ensure that approved firms are legitimate businesses with real operations. Without stricter oversight, the government risks allowing high-end AI chips to move into the hands of prohibited organizations through these gaps.

The future of chip exports

Moving forward, the government will need to address how it verifies the legitimacy of Chinese firms. The New Matrix case provides a blueprint for how shell companies can exploit current rules. Future vetting may need to include more rigorous physical inspections and deeper ownership audits to prevent such occurrences.

Nvidia will likely face continued scrutiny as it balances its commercial interests with national security requirements. The company must navigate a landscape where trade rules are constantly evolving. Any further lapses in oversight could lead to stricter regulations or a reversal of the current relaxed trade environment.

The situation highlights the ongoing tension between economic interests and security concerns. As AI technology becomes more critical, the methods used to control its distribution will remain a primary focus for policymakers. The oversight of these sales will be a key indicator of how effectively the US can manage its technological competition with China.

What to watch

Sources and links

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

Listen to the AI Wire headlines

Listen to the AI Wire

Listen to our daily AI shows every morning: on your commute, in the shower, wherever you are

Listen to the daily shows

First reported by The Verge · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

More AI news