Microsoft and Amazon have expressed support for U.S. legislation that would restrict the export of advanced AI chips to China, a move that highlights growing concerns over national security and global technology competition.
According to unnamed sources, Amazon’s cloud division has privately conveyed backing to Senate staff, while Microsoft has publicly endorsed the initiative.
The bill, known as the GAIN AI Act, is part of the broader National Defense Authorization Act (NDAA) and would require AI chipmakers, including Nvidia, to prioritize U.S. orders for cutting-edge processors before fulfilling international requests.
The GAIN AI Act introduces a “right of first refusal” for domestic buyers, giving U.S. companies 15 days to secure advanced GPUs before overseas shipments.
This framework could give companies like Microsoft and Amazon preferential access to Nvidia’s latest technology, particularly as global supply tightens.
Industry insiders warn that the law may reduce international competition for AI chips and limit computing power availability outside the U.S., potentially creating disparities in AI development capabilities globally. Companies such as Meta and Google have yet to take a formal position on the legislation.
Washington has been increasingly focused on limiting China’s access to high-end AI hardware over security concerns. Lawmakers argue that advanced GPUs could be used in applications with potential national security implications, including defense-related AI research.
Amazon and Microsoft’s support is widely interpreted as a strategic move to secure supply in an increasingly competitive market.
With U.S. private AI investment reaching $470 billion from 2013 to 2024, the demand for high-performance computing continues to outpace supply, making prioritized access to top-tier chips highly valuable.
If enacted, the GAIN AI Act could reshape global AI infrastructure. Non-U.S. companies may face tighter GPU access, prompting foreign cloud providers to establish U.S.-based operations to secure high-end processors.
Colocation operators, which rent data center space and power, could become key intermediaries, allowing AI labs and enterprises abroad to access U.S. GPUs indirectly.
Data center capacity trends illustrate the gap in compute resources. In 2024, the EU added 1.6 gigawatts (GW) of data center capacity compared to 5.8 GW in the U.S., leaving substantial room for U.S.-hosted services to cater to international clients. Analysts predict AI data center power requirements could reach 200 gigawatts by 2030, further intensifying demand for advanced chips
The post Microsoft and Amazon Back AI Chip Export Restrictions Amid China Security Concerns appeared first on CoinCentral.


