To Beat China, America Must Balance AI Innovation with National Security

While the White House is busy restricting trade with China, it keeps unlocking the back gate. Even as the Department of Commerce expanded export restrictions on March 26 to prevent Chinese firms from accessing American artificial intelligence, the White House is welcoming Emirati dollars to build out U.S. AI infrastructure, despite close ties between Emirati and Chinese technology firms.

Amid Washington’s efforts to stymie Chinese AI development, these ties merit scrutiny to ensure that America’s technology partners share its assessments of the risk of adversarial AI development.

Plans to Build U.S. AI Infrastructure

On March 20, UAE National Security Advisor Sheikh Tahnoon bin Zayed and a delegation of heads of sovereign wealth funds and corporations traveled to Washington. Alongside President Trump, Zayed committed to a decade-long, $1.4 trillion investment framework to fund the construction of next-generation data centers, semiconductor manufacturing, and energy infrastructure.

Concurrently, UAE companies announced new partnerships and reiterated support for existing collaboration. Abu Dhabi-based investment group MGX, for example, welcomed NVIDIA and xAI to its existing collaboration with Microsoft, BlackRock, and Global Infrastructure Partners. Dubbed the AI Infrastructure Partnership and formed last year, the initiative aims to support global AI supply chains via joint financing. UAE investment fund ADQ and the U.S.-based Energy Capital Partners announced a separate $25 billion investment in energy infrastructure and data centers.

The deal is the latest in the administration’s sweeping bid to promote investment in U.S. AI infrastructure. Just two days after his inauguration, President Trump announced a $500 billion joint venture called Stargate, funded by OpenAI, Oracle, and SoftBank, to encourage innovation by building data centers and energy infrastructure in Texas. 

UAE Tech Enmeshed with China

The new and expanded U.S.-UAE partnerships come amid persistent concerns about Chinese investment in the UAE’s technology sector. While Microsoft and BlackRock have their own problematic ties with China, the relationships that have raised the most red flags are OpenAI’s and Microsoft’s links with G42, the jewel in the crown of UAE’s AI development. 

G42 retains close ties to China’s military industrial complex and U.S.-blacklisted Chinese companies, including Huawei and the Beijing Genomics Institute. Most concerningly, last year, the House Select Committee on the Chinese Communist Party accused G42’s chief executive, Peng Xiao, of operating Chinese and Emirati companies that “materially support” China’s military development and human rights abuses through the development of dual-use technologies. 

After this scrutiny from lawmakers, Microsoft pared back its agreement with G42 to include only the lease of its AI products rather than sensitive technology transfer. G42, on the other hand, made no changes to address the concern about the firm’s ability to share American technology with China. Beijing’s long history of stealing strategically important intellectual property pales in comparison to China’s potential to amass U.S. technology through willing partners around the world.

Washington’s Efforts to Stem Chinese AI Advancement

U.S. efforts to undermine China’s use of AI to modernize and expand its military capabilities have largely remained separate from efforts to expand U.S. AI innovation. Most recently, in its March 26 announcement, the Commerce Department on March 26 added over 50 Chinese firms allegedly involved in military activity to its Entity List, restricting their capacity to purchase or re-export U.S. technologies.

Commerce targeted Chinese technology firms accused of developing AI models and other components for the People’s Liberation Army (PLA). The banned entities include subsidiaries of Inspur Group, China’s largest data and computing server manufacturer and a key customer of Western chipmakers such as NVIDIA and Intel. In its announcement, Commerce accused the firms, some of which had American clients, of contributing to the construction of a military supercomputer and aiding the development of hypersonic weapons. Commerce also designated the Beijing Academy of Artificial Intelligence for developing AI models and high-end chips for military use.

The most recent announcement belies the connections between Chinese technology companies and the PLA. Beijing’s policy of military-civil fusion (MCF) seeks to integrate China’s civilian economy into the country’s military industrial base, allowing the PLA to rapidly weaponize cutting-edge innovations. 

The Entity List additions are the latest in a series of efforts to thwart MCF and prevent China’s weaponization of U.S. technology. Beginning in 2022, the Biden administration introducedincreasingly stringent export controls on semiconductors and other related components, culminating in the release of the Framework for AI Diffusion. This framework, intended to prevent countries from diverting computing components to China, tightened export controls on semiconductors while limiting the spread of powerful AI models.

U.S. Must Strengthen Export Controls and Invest in Domestic AI Capabilities

The Trump administration cannot have it both ways. Shutting the front door for Chinese companies while simultaneously opening the back gate for questionable inbound investment is antithetical to Washington’s efforts to win in its rivalry with the Chinese Communist Party (CCP).

Continuing to add Chinese firms to the Entity List is essential to national security, but a company-by-company approach to Chinese MCF will remain a game of whack-a-mole as the CCP backstops banned companies with other state-backed entities. The Trump administration should take an industry-based approach to undermining Beijing’s technology-enabled military modernization. By finalizing the Framework for AI Diffusion, the administration can undermine Beijing’s military modernization efforts, cut off backdoor access to American innovations, and strike a blow against the CCP’s economic model, achieving three goals for the price of one. Congress, meanwhile, should advance proposed measures to limit American investment into the Chinese technology sector.

Meanwhile, the Trump administration should work to harmonize this sector-based approach with U.S. partners and allies to align technology export controls so that China — and other U.S. adversaries — cannot circumvent U.S. restrictions. Future technology collaboration between Washington and its allies should ensure that partners share the same alarm about the CCP.

Johanna “Jo” Yang is a research and editorial associate at the Center on Cyber and Technology Innovation at the Foundation for Defense of Democracies. Jack Burnham is a research analyst at the Foundation for Defense of Democracies.