By Ana Swanson
Earlier this week, the Biden administration unveiled extensive regulations on the sharing of artificial intelligence chips and models with foreign nations, seeking to create a global framework that will influence the manner in which AI disseminates globally in the ensuing years.
As the influence of AI expands swiftly, the Biden administration argued that these regulations are essential to maintain control over this revolutionary technology for the benefit of the United States and its allies, while preventing adversaries from utilizing it to enhance their military capabilities, conduct cyber warfare, and pose threats to national security.
Technology firms have expressed their discontent with the new regulations, claiming they jeopardize their sales and the future viability of the U.S. tech sector.
The regulations impose various restrictions on the quantity of AI chips that companies are permitted to export to different nations, effectively categorizing the world into three groups. The United States and 18 of its closest allies—such as the United Kingdom, Canada, Germany, Japan, South Korea, and Taiwan—are exempt from any constraints and can purchase AI chips without limitations.
Nations currently under U.S. arms embargoes, including China and Russia, will continue to encounter existing prohibitions on AI chip acquisitions.
All remaining nations—covering most of the globe—will be subjected to import limits on AI chips; however, countries and firms may be able to elevate that limit through special agreements with the U.S. government. These regulations may cause dissatisfaction among some foreign governments: Countries that are close trade partners or military allies of the U.S., such as Mexico, Switzerland, Poland, and Israel, will also encounter limitations on their capacity to acquire substantial quantities of American AI products.
In a statement, the European Commission expressed concerns about the policies when it communicated with the Biden administration.
“We are worried about the U.S. policies enacted today that limit access to advanced AI chip exports for certain EU member states and their businesses,” the commission stated, adding that European nations “represent an economic opportunity for the U.S., rather than a security threat.”
The objective of the regulations is to prevent China from sourcing the technology necessary for producing AI from other countries, after the U.S. curtailed such transactions with China in recent years.
However, the regulations also pursue broader objectives: they aim to position allied nations as prime locations for companies building the biggest data centers globally, ensuring the most advanced AI models remain within the United States and its allies.
Around the world, particularly in the Middle East, governments have been investing significantly in attracting and constructing vast data centers to establish themselves as the next hub for AI development.
Jake Sullivan, President Joe Biden’s national security adviser, informed reporters on Sunday that the regulation would guarantee that the infrastructure tasked with training the most advanced AI resides in the U.S. or in territories belonging to close allies, preventing the offshoring of such crucial capacities like chips and batteries—industries that have required significant investment to bring back onshore.
Sullivan explained that the regulation would provide “greater clarity to our international partners and to industry,” while simultaneously addressing national security concerns posed by malicious actors who might exploit “American technologies against us.”
The determination of whether to uphold or enforce the new regulations will rest with the Trump administration. During a call with reporters on Sunday, Biden administration officials asserted that the regulations enjoyed bipartisan backing and that they had engaged in discussions with the incoming administration regarding them.
Despite Chinese companies beginning to develop their own AI chips, the global market for these semiconductors is predominantly controlled by U.S. firms, notably Nvidia. This control has granted the U.S. government the power to manage the flow of AI technology worldwide by imposing restrictions on exports from American companies.
Firms have protested against these restrictions, asserting that the limitations could hinder benign or even advantageous types of computing, irritate U.S. allies, and ultimately drive global buyers toward non-American alternatives, such as those produced by China.
Ned Finkle, Nvidia’s vice president for government relations, labeled the regulation “unprecedented and misguided,” asserting that it “threatens to obstruct innovation and economic advancement globally.”
“Instead of alleviating any threat, the new Biden regulations would merely weaken America’s competitive edge on the global stage, undermining the innovation that has maintained the U.S. at the forefront,” he articulated. Nvidia’s stock fell nearly 2% on Monday.
Brad Smith, the president of Microsoft, expressed confidence that the company would be able to “fully comply with this regulation’s elevated security standards and address the technology demands of nations and clients around the world that depend on us.”
In a letter to congressional leadership on Sunday that was reviewed by The New York Times, Jason Oxman, president of the Information Technology Industry Council, which represents tech companies, requested Congress to intervene and exercise its authority to overturn the action if the Trump administration refrains from doing so.
John Neuffer, the president of the Semiconductor Industry Association, expressed his group’s “deep disappointment” regarding the hasty execution of a policy shift of this scale and significance just days before a presidential transition and without adequate input from industry representatives.
“The stakes are high, and the timing is delicate,” Neuffer noted.
The regulations, encompassing over 200 pages, establish a system where companies operating data centers, like Microsoft and Google, can apply for special governmental accreditations.
In exchange for adhering to specific security protocols, these companies may trade AI chips more freely across the globe. However, they are still required to maintain 75% of their total AI computing capacity within the United States or allied countries and limit their computing power in any single foreign nation to no more than 7%.
Furthermore, the regulations initiate the first controls over weights for AI models, which are the parameters specific to each model that dictate how AI makes predictions. Firms establishing data centers overseas will be mandated to adopt security standards to safeguard this intellectual property and inhibit adversaries from acquiring access to it.
Governments facing restrictions can elevate their import limits on AI chips by signing agreements with the U.S. government, committing to align with U.S. objectives regarding the protection of AI.
Last year, under U.S. government supervision, Microsoft finalized an agreement to partner with the Emirati firm G42, which involved G42 removing Huawei apparatus from its systems and undertaking other measures.
