New Filings Put Restricted Nvidia AI Chips Back in the Spotlight Amid China Export-Control Scrutiny
Fresh reporting has added another layer to the widening controversy over restricted Nvidia AI chips and their alleged movement into China, raising new questions about data-center financing, server ownership, and global enforcement of U.S. export controls.
According to a Bloomberg report citing Chinese regulatory filings, a holding company connected to Chinese executive Alice Huang disclosed restricted Nvidia GPUs as part of equipment used for financing collateral in 2025. The disclosure reportedly involved servers linked to a Shanghai data-center operation and comes amid broader investigations into how high-end AI hardware may have reached China despite U.S. restrictions.
The company named in the filings, Shanghai Longteng Intelligent Computing Data Group Co., reportedly gained access to servers originally purchased by Aperia Cloud from Dell. Aperia Cloud’s chief executive has faced fraud-related charges tied to the alleged smuggling of Nvidia GPUs into China.
A unit of Longteng, identified as Shanghai Longjie Yaoteng, later listed those servers among equipment pledged as collateral for financing, according to the report. Bloomberg also noted that Alice Huang founded Longteng in 2016 and gave up her shares in 2017.
The details are significant because Huang has already been linked in reports to Megaspeed, a company previously examined over claims involving the supply of restricted Nvidia AI chips to China. While U.S. intelligence officials are reportedly not convinced that Huang was acting as a spy, the filings appear to add to concerns about how advanced AI chips may have been routed through complex corporate structures and overseas supply chains.
The latest revelations come at a sensitive time for the semiconductor industry. U.S. export controls have targeted Nvidia’s most powerful AI accelerators, including chips such as the H100 and H200, due to concerns that they could support advanced artificial intelligence development with military or strategic applications.
Bloomberg also reported that servers made by Super Micro and equipped with Nvidia H100 or H200 chips appeared in the regulatory disclosure. Super Micro has denied making such direct sales, stating that it follows U.S. law and company policy. The company has also said that individuals previously accused of diverting restricted chips acted against its rules.
In March, U.S. authorities charged several people tied to Super Micro-related operations over alleged efforts to move restricted GPU technology to China. The case intensified attention on whether high-performance AI servers can be diverted after purchase through resellers, shell companies, or third-party logistics networks.
The issue is not limited to the United States and China. Malaysia is now reviewing its legal framework to determine whether it can better address the re-export of GPUs. The country’s Minister of Investment, Trade and Industry, Datuk Seri Johari Abdul Ghani, said officials are evaluating current laws as part of efforts to align with U.S. sanctions and prevent Malaysia from being used as a transit point for restricted AI hardware.
At the same time, U.S. lawmakers are pushing for tougher rules designed to stop Chinese artificial intelligence firms from accessing advanced chips indirectly. One concern is remote access, where companies may use powerful GPUs hosted in overseas data centers rather than importing the hardware directly.
That loophole has become a growing focus as analysts examine whether traces of restricted Nvidia Blackwell-class GPUs appear in the development or operation of leading Chinese AI models. If confirmed, such findings could strengthen arguments for tighter monitoring of global cloud computing infrastructure and AI data centers.
The broader dispute highlights the growing importance of GPUs in the global technology race. Nvidia’s AI chips are considered essential for training and running advanced artificial intelligence systems, making them one of the most strategically important technologies in the world. As demand continues to surge, regulators are increasingly focused on where these chips are sold, how they are financed, and who ultimately controls access to them.
For Nvidia and its hardware partners, the controversy underscores the challenges of managing compliance across a sprawling international supply chain. Even when manufacturers follow export-control rules at the point of sale, chips and servers can still move through intermediaries, financing arrangements, leasing structures, and overseas data-center operators.
The new filings do not settle the larger questions surrounding restricted AI chips in China, but they add to the growing pressure on governments, chipmakers, server manufacturers, and cloud providers to strengthen oversight. As artificial intelligence becomes more central to economic and national security strategy, the battle over advanced GPUs is likely to remain one of the most closely watched technology issues worldwide.
