The proposed Remote Access Security Act (RASA), currently under consideration in both the U.S. House of Representatives and the Senate, aims to close the so-called cloud compute loophole. By authorizing the U.S. government to regulate not only the export of AI capabilities but also their usage, the legislation seeks to prevent adversarial nations from accessing U.S. technologies to train AI models. While this appears to be a logical extension of existing hardware and software export controls, experts warn that it may inadvertently deliver a competitive advantage to Chinese cloud operators, ultimately displacing U.S. technology globally.
For several years, reports have suggested that China has utilized cloud infrastructure, particularly in Singapore and other Southeast Asian nations, to circumvent U.S. export bans on advanced AI chips. Since early 2024, the United States has moved to address this by proposing requirements for cloud companies to identify whether data centers are accessed by companies from adversarial nations. By January 2025, the Biden administration’s Framework for Artificial Intelligence Diffusion established a system for validated end-user authorizations, functioning similarly to a “know your customer” requirement for cloud operators.
The policy landscape has shifted rapidly over the past two years. In April 2025, the Trump administration tightened controls on Nvidia’s H20 chips, which were designed to comply with previous restrictions. Industry leaders warned that these measures would accelerate the adoption of domestic Chinese alternatives, such as those produced by Huawei. By late 2025, the administration reversed course, approving H20 exports and eventually the more capable H2000 chips under a licensing regime.
Critics of the proposed RASA legislation argue that while restricting the most advanced technologies to adversaries is prudent, cloud services and AI chips are distinct entities. They contend that heavy-handed restrictions on global cloud providers like Amazon Web Services, Microsoft Azure, Google Cloud, and facility operators such as Equinix and Digital Realty could prove devastating. If these companies are forced to demand that global customers “prove their innocence” rather than simply denying services to a defined blacklist, the resulting uncertainty may drive legitimate users toward Chinese providers like Huawei Cloud and AliCloud.
This shift would contradict the goals set forth in President Donald Trump’s Executive Order 14320 and his “America’s AI Action Plan,” which prioritize the global export of the American AI technology stack. China’s long-term objective is to foster indigenous firms that can displace American leaders globally. Therefore, U.S. policymakers are urged to exercise caution; implementing overly restrictive measures that slow down U.S. firms may ultimately help Beijing achieve its goal of domestic and international technological dominance. Maintaining an open global market remains essential for ensuring that the world continues to choose American technology over Chinese alternatives.
