SIAgentNews.com
← All news
DevelopingChips

U.S. House releases data on Nvidia and AMD AI chip export licenses to China

New data reveals that while millions of AI chip applications were filed, the majority of approved licenses remain unshipped.

· First reported by KuCoin

3.44 millionTotal AI chip applications

The short version

  • Companies applied for 3.44 million AI chips, with ByteDance and Alibaba accounting for nearly 2 million of those requests.
  • The Bureau of Industry and Security (BIS) approved 879,000 licenses under the Export Control Reform Act.
  • Most of the approved chips are currently in a 'not yet shipped' status.
  • The data includes licenses for companies designated by the U.S. Department of Defense as Chinese military enterprises.
  • Nvidia reportedly applied for licenses to assist sales to China through potential smuggling channels.

Why it matters

This disclosure provides a rare look at the scale of hardware restrictions and how major tech firms are navigating trade barriers. It highlights the tension between commercial demand for AI hardware and national security concerns regarding military use.

A look at the export data

Brian Mast, who leads the U.S. House Foreign Affairs Committee, recently shared a dataset concerning the shipment of AI chips to China. This information tracks hardware from major producers like Nvidia and AMD, specifically looking at permissions granted under the Export Control Reform Act. This release represents the most detailed public disclosure of this kind since the act was first implemented.

The data reveals a massive scale of interest from Chinese tech giants. According to the records, companies collectively submitted 3.44 million applications for AI chips. This volume demonstrates the intense demand for high-end computing power required to train and run large-scale artificial intelligence models in the region.

Out of those millions of requests, the Bureau of Industry and Security (BIS) granted approval for 879,000 licenses. While this is a significant number of permissions, it represents only a fraction of the total applications submitted by the companies. The gap between requests and approvals highlights the strict scrutiny the U.S. government applies to high-end technology transfers.

Major tech firms and volume

The data identifies specific heavy hitters among the applicants. ByteDance submitted the highest number of applications, totaling 1.015 million chips. Alibaba followed closely behind with 949,000 applications. Together, these two companies accounted for nearly 2 million of the total requests found in the disclosure.

Other notable names appeared on the list, including Tencent, Didi, and Kuaishou. Interestingly, the gaming company MiHoYo received approval for approximately 57,800 chips. This amount was higher than the approvals received by other major tech entities like Baidu and JD.com.

The inclusion of these companies shows that the demand for AI hardware is spread across various sectors of the Chinese economy. From social media and ride-sharing to gaming and e-commerce, the need for specialized chips is a common thread across the country's primary technology platforms.

Shipping status and logistics

One of the most striking findings in the report is the current status of the approved hardware. The majority of the 879,000 approved chips are currently listed as 'not yet shipped.' This suggests that while the government has granted permission for the sales, the physical movement of the goods has not yet occurred for most of them.

The fact that these chips remain in a pending state raises questions about the liquidity of the hardware market. It also suggests that companies may be waiting for specific timing or logistical conditions to be met before moving the products. This delay could have implications for how quickly Chinese firms can scale their AI capabilities.

Furthermore, the data suggests that the approval process is not a simple green light for immediate delivery. The discrepancy between approval and shipment may reflect complex supply chain hurdles or the cautious approach of the manufacturers. It provides a snapshot of a bottleneck where regulatory permission meets real-world logistics.

Security and smuggling risks

The disclosure also highlights the security concerns driving these export controls. Some of the licenses involved exports to companies that the U.S. Department of Defense has designated as Chinese military enterprises. This confirms that the government is actively monitoring the flow of chips to prevent them from being used for defense purposes.

The documents also contain a more controversial detail regarding Nvidia. The records show that the company sought to help move products to China via possible illicit trade routes. This indicates that manufacturers are looking for creative ways to reach their customers despite the increasingly rigid trade barriers set by the U.S. government.

These findings underscore the ongoing tension between global trade and national security. As the U.S. continues to tighten its grip on high-end semiconductors, the data shows a complex landscape of compliance, evasion, and high-stakes competition for the underlying hardware that powers modern AI.

What to watch

  • Whether the 'not yet shipped' status changes as companies begin to move the approved hardware.
  • How the crypto market reacts to the potential impact of stricter export controls on hardware liquidity.
  • Future updates on Nvidia's efforts to navigate smuggling channels or other export workarounds.

Try it

Look up the current export restrictions on high-end GPUs to see which specific models are currently restricted for sale in China.

Sources and links

First reported by KuCoin. 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.

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.

More AI news