During a three-day diplomatic tour of the Middle East, President Trump and representatives from Silicon Valley have helped elevate the Persian Gulf from a newcomer to a key player in artificial intelligence.
They have finalized a significant agreement with the United Arab Emirates to supply hundreds of thousands of Nvidia's most advanced AI chips annually, aiming to establish one of the largest data center hubs worldwide in the region, according to sources familiar with the negotiations. Shipments are expected to commence this year, with about 100,000 chips designated for the Emirati AI company G42, and the remainder allocated to U.S.-based cloud service providers.
The administration announced the deal alongside plans for a new AI campus in Abu Dhabi, powered by 5 gigawatts of electricity. This facility would become the largest of its kind outside the U.S., facilitating American companies' service to customers across Africa, Europe, and Asia. However, the specifics regarding the chip technology remain undisclosed and may still be subject to revision.
In parallel, the U.S. has secured multibillion-dollar agreements to export advanced chips from Nvidia and AMD to Saudi Arabia. Negotiations are ongoing over a broader AI technology contract with the Saudi government, according to insiders.
These AI agreements have sparked debate within and beyond the administration over whether the eagerness to forge deals in a region with financial ties to the Trump family risks outsourcing America's future technology sector.
The discussions highlight internal divisions on AI policy. The deals were brokered in the Middle East by David Sacks, the administration’s AI coordinator, and Sriram Krishnan, senior AI policy adviser—both veteran venture capitalists. Influential AI industry leaders, including OpenAI's Sam Altman and Nvidia's Jensen Huang, have also participated in ongoing discussions concurrent with the president’s visit.
These officials advocate for building a network of partnerships among companies and countries to promote mutual interests, confident that such deals will strengthen American AI firms and expand the nation's lead in the field.
“Our goal is for American AI to proliferate globally,” said Sriram Krishnan.
However, as details reached Washington, concerns emerged from current and former officials about potential shortcomings in safeguards preventing the technology from benefiting China. Some worry that by the decade’s end, the world’s largest data centers may be situated in the Middle East rather than the U.S.
These tensions underscore contradictions in President Trump’s approach. While he has criticized offshoring of U.S. industries and recently imposed tariffs to reverse the trend, his first major foreign trip revealed a willingness to support American AI companies expanding overseas.
The announcements coincide with reports of over $2 billion flowing recently to Trump-affiliated companies from Middle Eastern sources, including Saudi-backed investments in Trump’s cryptocurrency ventures and plans for a new presidential aircraft from Qatar.
Klon Kitchen, a nonresident senior fellow at the American Enterprise Institute, acknowledged the administration’s aim to ensure “American technologies form the foundation for all AI,” while cautioning about the risks of exporting AI development overseas, similar to past patterns in the energy sector.
“In foreign policy, trade-offs are often unavoidable,” Kitchen noted.
The U.S. has maintained global leadership in AI largely because its companies pioneered much of the underlying technology. Worldwide demand for Nvidia chips and partnerships with American AI firms is surging, especially from the UAE and Saudi Arabia, which seek to diversify their economies away from oil dependency.
However, the U.S. government must authorize these foreign chip sales to prevent sensitive technology—potentially applicable to military uses and autonomous weapons—from reaching adversaries like China.
The previous administration approved some chip sales to the UAE but hesitated to meet all demands, fearing such deals could eventually enable the UAE to surpass the U.S. in AI capabilities. Concerns also centered on providing advanced technology to an authoritarian regime with close ties to China.
In 2024, Tahnoun bin Zayed Al Nahyan, chairman of G42 and UAE national security adviser, proposed a plan to build a 5-gigawatt computing cluster in the UAE, similar to what the Trump administration later approved. However, the prior administration rejected the plan over worries it would offshore AI jobs and critical security infrastructure.
Following President Trump’s inauguration, advisers feared that continued restrictions might drive the UAE toward Chinese alternatives, especially as Huawei's AI chips improve, though they have not yet been exported.
“The president challenged us to win the AI race,” said David Sacks at a Saudi Arabia conference. “We need strategic partners like Saudi Arabia to build on our technology.”
Negotiators from the UAE and Saudi Arabia proposed using U.S. chips and AI models to expand American companies’ regional reach, while offering security guarantees about the chips’ location and usage.
U.S. officials also pushed for reciprocal investments, requiring that every data center built in the Middle East for a U.S. firm be balanced by financial support for a facility in the United States. The UAE has agreed to this arrangement.
Additionally, the administration announced efforts to streamline UAE investments in the U.S. and establish a working group to oversee this process.
The full terms of these agreements remain unclear, and companies like G42 still require licensing and approvals to receive chip shipments.
Some AI experts sympathize with views within the administration that the U.S. lacks sufficient energy resources to build all the data centers global demand requires. These limitations have constrained companies such as OpenAI, which recently restricted a popular animated image feature due to capacity issues.
J.J. Kardwell, CEO of data center provider Vultr, explained that U.S. firms are increasingly considering overseas data centers because domestic power availability this year supports only about 25,000 of Nvidia’s newest AI chips, whereas Gulf states have abundant energy to quickly support 100,000 chips or more.
“The U.S. cannot be the sole data center provider globally,” Kardwell said.
However, these perspectives are not universally accepted within the Trump administration, which otherwise emphasizes an “America First” agenda focused on boosting domestic energy production and reshoring industries. Some officials accuse Sacks and Krishnan of independently negotiating deals with Middle Eastern governments without broader consensus.
Critics of offshoring argue that each overseas data center means fewer American jobs in construction and related trades, as well as diminished tax revenues and reduced U.S. influence over AI's economic and military benefits.
Jimmy Goodrich, senior adviser at RAND for technology analysis, described establishing AI data centers abroad as “taking the easy way out.” While he recognizes frustrations with permitting and the lack of nuclear power domestically, he insists the solution is not to relocate facilities to the Middle East.
Goodrich emphasized that the U.S. still leads in AI talent, companies, and hardware, and should focus on accelerating permitting processes and improving its energy infrastructure to maintain that edge. Building major data centers in the Gulf risks empowering regional or Chinese competitors.
“This is a scenario we have witnessed before and should avoid repeating,” he warned.
Sam Winter-Levy, a fellow at the Carnegie Endowment for International Peace, stated that the large chip sales appear inconsistent with an “America First” AI or industrial policy.
“Why would we outsource the infrastructure underpinning the key industrial technology of the coming years?” he questioned.
0 Comments
No comments yet. Be the first to comment!