In a move aimed at facilitating the global development and supply of artificial intelligence (AI), the US Commerce Department has brought a new policy that will ease export restrictions for advanced AI chips to data centers in the Middle East and Central Asia.
This move could simplify the shipment process for US companies by allowing foreign data centers to apply for special status, avoiding the need for individual export licenses.
Earlier, the US exporters were required to obtain individual licenses to export such chips to “countries of concern.”
The new rule introduces the “Validated End User” (VEU) program, allowing foreign data centers to receive AI chips like those produced by Nvidia under a general authorization, bypassing the need for US companies to seek individual export licenses.
“This update to the VEU program was designed to protect national security by ensuring high standards for physical and cybersecurity at data centers that house advanced AI systems,” a US Commerce Department statement said. “It will also reduce licensing burdens on industry by allowing data centers to fulfill the stringent requirements of the VEU program upfront, enabling US exporters to ship designated items to pre-approved entities under a general authorization, instead of under multiple individual export licenses.”
This development comes after the US imposed stringent licensing requirements in October 2023, restricting the shipment of advanced AI technology to regions including the Middle East and Central Asia.
Concerns over the potential misuse of these technologies, particularly the risk of them being funneled to China through third-party countries, have driven these export controls.
“Through the VEU program, we will collaborate with foreign data centers and their respective governments to ensure that US technology is safeguarded from misuse,” the statement added.
The move is seen as a response to growing pressure from US companies, especially in the AI chip sector. The rule, however, provides a clear pathway for data centers to access advanced AI chips more efficiently while addressing security concerns raised by US lawmakers.
Nvidia declined to comment on this matter.
Addressing the concerns
One major concern in Washington is the possibility that AI technology could find its way to China via third-party entities in the Middle East. G42, a UAE-based AI company with former ties to China, has been under scrutiny.
Despite its claims of severing Chinese connections and complying with US restrictions, the company’s $1.5 billion deal with Microsoft earlier this year raised alarms, especially among China hawks in Congress.
Although the new rule addresses the concerns, it won’t be a cakewalk for data center operators such as G42.
The VEU program will require data centers applying for the status to undergo a rigorous vetting process. The review will include assessments of their customer base, business activities, cybersecurity protocols, and access controls. Approved entities will be subject to strict reporting requirements and on-site inspections by US officials to ensure compliance.
“The Data Center VEU program will rigorously vet applicants to ensure that any authorization includes appropriate safeguards and security measures that protect our most advanced technologies,” Alan F Estevez, Under Secretary of Commerce for Industry and Security, said in a statement.
Furthermore, host governments may need to offer guarantees regarding the secure and appropriate use of these technologies.
The Commerce Department emphasized that each authorization will be limited, specifying the types and quantities of AI technology that can be exported to ensure that US national security is protected.
“Our goal is to strike a balance between supporting international AI innovation and mitigating the risks to US security,” Estevez added.
This rule comes at a time when the US is grappling with how best to control the flow of advanced technologies without stifling innovation, both domestically and internationally. The decision to implement a VEU program is expected to ease some of the tensions between the tech industry and Washington while maintaining oversight on how AI technologies are utilized abroad.
Source:: Network World