Deutsch
English
Español
Français
Nederlands
Português
On May 31, 2026, the US Commerce Department's Bureau of Industry and Security (BIS) closed a year-long loophole that allowed Chinese AI companies to acquire advanced semiconductors—including Nvidia's Blackwell and Rubin processors and AMD's MI350x—through overseas subsidiaries in Malaysia, Singapore, and the UAE. Industry sources estimate hundreds of thousands of chips, potentially worth billions of dollars, flowed through this gap since enforcement of the AI Diffusion Rule was paused in May 2025. This article analyzes the strategic implications: how the regulatory gap emerged, what it reveals about the limits of unilateral export controls, and why both Washington and Beijing now face a new phase of technological confrontation where no side has achieved a decisive advantage.
How the Loophole Emerged
The origins of the loophole trace back to May 2025, when the Trump administration announced it would not enforce the Biden-era AI Diffusion Rule. That rule had imposed strict licensing requirements on advanced chip exports to Chinese entities, but the enforcement pause created ambiguity: did the restrictions apply to overseas subsidiaries of Chinese firms? Without clear guidance, Chinese AI companies—including ByteDance, Alibaba, and Baidu—began routing orders through subsidiaries in Malaysia, Singapore, and the UAE. These entities were legally incorporated in third countries and could purchase Nvidia's Blackwell GB200 and AMD's MI350x chips through normal commercial channels. The US semiconductor export controls had a gap large enough to drive a truck through.
The Scale of the Evasion
BIS estimates that hundreds of thousands of servers—potentially millions of GPU-equivalent compute units—flowed through this gap. One industry source told CNBC that the number of chips exported could be in the hundreds of thousands. At current market prices, a single Nvidia Blackwell B200 GPU costs around $30,000–$40,000, meaning the total value of chips that evaded controls could exceed $100 billion. This is not a minor leak but a flood that directly challenges the effectiveness of US technology containment strategy. The AI chip export ban impact on US national security objectives is now under intense scrutiny.
The New BIS Guidance: Closing the Side Door
The guidance issued on May 31, 2026, shifts from a geography-based enforcement model to a beneficial ownership standard, similar to OFAC financial sanctions. Under the new rules, licensing requirements apply to any entity whose ultimate parent or headquarters is in China, regardless of where the subsidiary is physically located or incorporated. This means a Chinese AI company's subsidiary in Kuala Lumpur can no longer purchase advanced chips without a license—which is typically denied. The guidance does not create new law but clarifies existing restrictions under the Export Administration Regulations (EAR).
What the Guidance Does and Does Not Do
The BIS guidance is carefully calibrated. It requires license enforcement for new shipments of advanced chips to China-linked entities anywhere in the world. However, it does not require data centers to stop using chips already acquired, nor does it cut off servicing of existing advanced computing equipment. This grandfathering provision means that the hundreds of thousands of chips already installed in Chinese AI data centers remain in use. The China AI chip stockpile built during the loophole period will continue to power Chinese AI development for years.
Strategic Implications: The Limits of Unilateral Export Controls
The existence and size of this loophole directly challenge the effectiveness of US technology containment strategy. Despite two years of escalating controls, Chinese AI firms have continued to access cutting-edge hardware. The loophole's closure comes too late to prevent a massive transfer of advanced computing power to China. Moreover, the episode has accelerated China's domestic chip development. Companies like Huawei have ramped up production of the Ascend 950 Pro, and Chinese AI labs have demonstrated competitive open-weight models. The US-China tech war 2026 is entering a new phase where neither side holds a decisive advantage.
Impact on Global Supply Chains
The new guidance also creates compliance challenges for cloud providers like AWS, Google Cloud, and Microsoft Azure. Their APAC customers with Chinese ownership structures are now affected, potentially disrupting cloud services. Countries like Malaysia, Singapore, and the UAE, which served as transshipment hubs, now face pressure to align with US export controls or risk secondary sanctions. The global semiconductor supply chain disruption is likely to intensify as companies scramble to verify beneficial ownership of their customers.
Expert Perspectives
This is the most significant escalation in US-China semiconductor controls since the original rules, said Sophie Turner, a geopolitical analyst. The loophole's existence and size directly challenge the effectiveness of US technology containment strategy, with consequences for AI development, global supply chains, and national security that are unfolding in real time. Legal experts at Greenberg Traurig noted that the guidance reinforces that the enforcement pause did not provide blanket relief—companies must continue to comply with EAR requirements.
FAQ
What exactly did the BIS guidance on May 31, 2026, change?
The guidance clarified that export licensing requirements for advanced AI chips apply to any entity with a Chinese parent or headquarters, regardless of where the subsidiary is located. This closed a loophole that had allowed Chinese firms to purchase chips through overseas subsidiaries in Malaysia, Singapore, and the UAE.
How many chips evaded the controls?
Industry sources estimate hundreds of thousands of advanced chips—potentially millions of GPU-equivalent compute units—flowed through the loophole between May 2025 and May 2026. The total value could exceed $100 billion.
Does the guidance affect chips already installed in data centers?
No. The guidance does not require data centers to stop using chips already acquired or cut off servicing of existing advanced computing equipment. The grandfathering provision means existing installations are unaffected.
What chips are affected?
The guidance targets Nvidia's Blackwell and Rubin processors, AMD's MI350x, and other advanced semiconductors on the Commerce Control List. Less advanced chips are not affected.
How have Chinese companies responded?
Chinese AI firms are now forced to either apply for export licenses (which are typically denied) or seek domestic alternatives such as Huawei's Ascend 950 Pro. The loophole closure is expected to accelerate China's push for semiconductor self-sufficiency.
Conclusion: A New Phase of Technological Confrontation
The closure of the $100 billion loophole marks a pivotal moment in US-China tech competition. While Washington has finally shut the side door, the damage is done: China has amassed a massive stockpile of advanced chips that will fuel its AI development for years. The episode reveals the inherent limits of unilateral export controls in a globalized economy. Moving forward, both Washington and Beijing face a new phase of technological confrontation where no side has achieved a decisive advantage. The race for AI supremacy is far from over—it has simply entered a more complex and contested chapter.
Follow Discussion