US tightens control over global AI chip exports

The U.S. government announced new measures on Monday, 13th January 2025, to further restrict exports of artificial intelligence (AI) chips and technology. The new regulations aim to ensure that advanced computing power remains within the U.S. and its allies, while limiting China’s access to these technologies.

The latest regulations impose a cap on the number of AI chips that can be exported to most countries, while granting unrestricted access to American AI technology for its closest allies. Exports to China, Russia, Iran, and North Korea will remain blocked. 

These new rules, introduced during the final days of President Joe Biden’s administration, aim to bolster American leadership in AI by controlling its development and distribution worldwide.

U.S. aims to maintain leadership in AI

“The U.S. leads AI now—both AI development and AI chip design—and it’s critical that we keep it that way,” said U.S. Commerce Secretary Gina Raimondo. The new regulations represent the culmination of a four-year effort by the Biden administration to limit China’s access to advanced AI chips that could enhance its military capabilities. By closing loopholes and imposing stricter controls, the U.S. hopes to preserve its dominance in AI.

The regulations will affect advanced graphics processing units (GPUs), which power data centers used to train AI models. Most of these chips are produced by Nvidia, based in Santa Clara, California, although other companies like Advanced Micro Devices also sell AI chips. 

Cloud service giants, such as Microsoft, Google, and Amazon, will be allowed to seek global authorizations for their data center projects, which will exempt them from the country quotas on AI chips.

Industry voices raise concerns

The new regulations have sparked criticism from industry leaders. Nvidia called the measures “sweeping overreach,” arguing that the White House is targeting technology already available in mainstream gaming PCs and consumer hardware. Oracle, a major data center provider, warned that the rules could hand the majority of the global AI and GPU market to Chinese competitors.

The regulations include worldwide licensing requirements for specific chips, with some exceptions. They also set controls on the “model weights” of advanced “closed-weight” AI models, which play a crucial role in decision-making for machine learning. 

The new measures categorize countries into three tiers: Tier 1 countries, which include Japan, Britain, South Korea, and the Netherlands, are exempt from most rules; approximately 120 other countries, such as Singapore, Israel, Saudi Arabia, and the UAE, will face export caps; and countries under arms embargo, like Russia, China, and Iran, will be fully banned from receiving the technology.

Global impact on AI computing power

U.S.-based companies like AWS and Microsoft, which are likely to receive global authorizations, will face restrictions on where they can deploy AI computing power. They will be allowed to use only 50% of their total AI computing power outside the U.S., with no more than 25% in Tier 1 countries and 7% in any single non-Tier 1 country.

The future of AI and its potential threats

While AI has the potential to revolutionize industries like healthcare, education, and food security, it also poses risks, including aiding in the development of biological weapons, supporting cyberattacks, and enabling surveillance and human rights violations.

“The U.S. has to be prepared for rapid increases in AI’s capability in the coming years, which could have transformative impacts on the economy and our national security,” said U.S. National Security Adviser Jake Sullivan.

These new regulations reflect the ongoing U.S. effort to control the global flow of AI technology and maintain its leadership in this rapidly advancing field.

Frequently asked questions (FAQs) about US AI chip export regulations

1. What are the new US regulations on AI chip exports? 

The US government has imposed stricter controls on the export of artificial intelligence (AI) chips and technology. These regulations limit the number of AI chips that can be exported to most countries, while allowing unrestricted access to US AI technology for its closest allies. Exports to countries like China, Russia, Iran, and North Korea will remain blocked.

2. Why is the US restricting AI chip exports? 

The US aims to maintain its leadership in AI development and chip design by controlling access to advanced computing power. The regulations seek to prevent China and other adversarial nations from using AI technology to enhance their military capabilities and to preserve US dominance in AI innovation.

3. Which countries are affected by these new regulations? 

The regulations divide the world into three tiers. Tier 1 countries, such as Japan, South Korea, and the UK, are largely exempt from the restrictions. About 120 other countries, including Singapore and Israel, will face export caps, while arms-embargoed nations like China, Russia, and Iran are fully banned from receiving the technology.

4. How do these regulations impact major companies like Nvidia and Microsoft? 

Major companies such as Nvidia, which manufactures AI chips, and cloud service providers like Microsoft, Google, and Amazon, will be subject to new licensing requirements and conditions for their data centers. Companies must comply with security measures, reporting demands, and human rights considerations to be eligible for global authorizations.

5. What are the restrictions on data center projects? 

Cloud service providers like Microsoft and Amazon can seek global authorizations to build data centers. However, these companies must follow specific conditions, including limits on the amount of AI computing power they can deploy outside the US and in non-Tier 1 countries. They can use only 50% of their total AI computing power outside the US, with no more than 25% in Tier 1 countries.

6. What is the significance of “model weights” in AI? 

Model weights are essential components of advanced AI models, helping to determine decision-making in machine learning. The regulations impose controls on the most advanced “closed-weight” AI models to prevent their misuse, particularly in military or surveillance applications.

7. How will these regulations affect US global competitiveness? 

While the regulations are designed to protect US technological supremacy, some industry leaders, like Nvidia, argue that these measures could hamper global innovation and give Chinese competitors an advantage. The US government, however, views these controls as necessary to safeguard national security and prevent the misuse of AI technology.

8. When do these new regulations take effect? 

The new regulations are set to take effect 120 days from their publication, giving the incoming administration time to review and potentially adjust the rules.

9. What are the potential risks of AI technology? 

While AI has significant benefits, such as improving healthcare, education, and food security, it also poses risks. AI could be used to develop biological weapons, support cyberattacks, and enhance surveillance systems that violate human rights. The US aims to address these threats by carefully controlling AI technology’s global distribution.

10. What are “Tier 1” countries, and why are they treated differently? 

Tier 1 countries, including the US’s closest allies like Japan, the UK, and South Korea, are largely exempt from the new restrictions. These nations are considered trusted partners in the development and deployment of AI technology, making them less likely to misuse the technology.