The Trump administration is contemplating a significant shift in U.S. technology export policy, potentially authorizing large-scale sales of advanced artificial intelligence chips to Gulf nations. This includes a potentially imminent deal with the United Arab Emirates’ AI firm G42 and an expected agreement with Saudi Arabia.
The move, under discussion ahead of President Trump’s visit to the Persian Gulf states this week, aims to strengthen U.S. ties in the region and boost tech exports, reports The New York Times. This approach diverges from the Biden administration’s stance, which had reportedly blocked similar sales due to concerns that providing powerful AI technology to autocratic governments with strong ties to China could erode the United States’ lead in AI development.
Negotiations are actively underway regarding a potential deal that could see hundreds of thousands of U.S.-designed AI chips sent to G42 Holdings. This Emirati AI firm has previously drawn scrutiny from the U.S. government over its connections to China. The discussions have reportedly generated internal tension within the Trump administration, pitting officials focused on business and technology interests against national security officials concerned about the potential misuse of the advanced technology.
David Sacks, the White House AI czar, has been involved in negotiating an agreement that would grant G42 access to chips with limited oversight, with some allocated for a partnership between G42 and OpenAI, and others going directly to G42.
Complementing the potential UAE deal, the Trump administration is also anticipated to announce an agreement this week with Saudi Arabia. This deal would provide the Saudi government and its recently launched AI company, Humain, with access to tens of thousands of semiconductors and technology support from leading AI chip manufacturers Nvidia and Advanced Micro Devices, according to The New York Times.
Saudi Arabia established Humain on May 12 as the primary vehicle for its ambitious AI strategy and investments. Chaired by Crown Prince Mohammed bin Salman and owned by the $940 billion Public Investment Fund (PIF), Humain intends to invest in and operate AI assets, including data centers, and prioritize the development of advanced large language models tailored for Arabic speakers.
The timing of Humain’s launch coincides with an expected U.S.-Saudi investment forum where significant deals in AI and other sectors are anticipated. Saudi Arabia plans a large server farm for AI processing with a $5 billion investment and an output of 1.5 gigawatts, highlighting its strategic need for graphics processors, Globes English reports.
Policy Overhaul and Industry Reactions
The potential sales to Gulf nations coincide with the Trump administration’s plans to overhaul a complex Biden-era regulation governing AI chip exports. The administration intends to rescind the ‘AI diffusion rule,’ which was scheduled to become effective on May 15. A Commerce Department spokesperson criticized the outgoing rule, stating “The Biden AI rule is overly complex, overly bureaucratic, and would stymie American innovation… We will be replacing it with a much simpler rule that unleashes American innovation and ensures American AI dominance.” The administration aims to replace it with a simpler, yet still robust, framework, potentially involving more targeted controls and bilateral agreements with nations seeking access to top-tier U.S. AI technology. This cancellation was influenced by pressure from countries like Saudi Arabia and the UAE, who felt discriminated against by the previous tiered system, according to Globes.
The debate surrounding the appropriate level of control has been highlighted by public disagreements within the industry. AI developer Anthropic, for instance, has strongly advocated for stricter controls, citing national security concerns. Anthropic even alleged sophisticated smuggling methods, such as “prosthetic baby bumps” and items “packed alongside live lobsters.”
In contrast, Nvidia has pushed back against expanding restrictions, arguing they could inadvertently push allies closer to China and harm U.S. innovation. A company spokesperson told CNBC that American firms should focus on innovation rather than telling “tall tales” about smuggling methods. Nvidia also emphasized China’s significant capabilities in AI, stating “China, with half of the world’s AI researchers, has highly capable AI experts at every layer of the AI stack. America cannot manipulate regulators to capture victory in AI,”. Cerebras CEO Andrew Feldman also called the Biden AI diffusion rule “bad policy.”
Geopolitical Landscape and Enforcement Challenges
Underlying the policy debate is the intensifying U.S.-China tech competition. U.S. export controls, while aimed at limiting China’s access to advanced AI capabilities, have simultaneously fueled China’s drive for semiconductor self-sufficiency. Domestic players like Huawei are rapidly advancing, preparing mass shipments of chips like the Ascend 910C and developing large-scale systems positioned as alternatives to restricted Western hardware. Analyst Paul Triolo suggested that Huawei’s Ascend 910C GPU will now become the hardware of choice for Chinese AI model developers.
The effectiveness of export controls also hinges on enforcement, a point underscored by concerns about sophisticated smuggling and reports regarding Huawei’s supply chain. While China’s SMIC foundry can produce 7nm-class chips, analysis suggests Huawei’s current advanced chips may utilize restricted components sourced through intermediaries.
This purported link has reportedly led to U.S. scrutiny of TSMC. The Trump administration is also seeking more aggressive measures, including tighter controls on AI chip exports and blocking servicing of key semiconductor equipment inside China. Officials have been in talks with Japan and the Netherlands to stop maintenance services for fabrication tools used by Chinese firms, as reported by Bloomberg.
G42’s Existing Partnerships and China Ties
G42, the Emirati firm at the center of potential chip sales, has existing significant partnerships with U.S. tech companies. Microsoft announced a $1.5 billion investment in G42 in April 2024, granting Microsoft minority ownership and a board seat.
This partnership involves G42 migrating its infrastructure to Microsoft’s Azure cloud services and a joint $1 billion developer fund. G42 has also partnered with U.S. AI chipmaker Cerebras Systems, unveiling the Condor Galaxy network of supercomputers. The first supercomputer in this network, CG-1, is located in Santa Clara, California.
G42 has been training its advanced AI models in data centers in California and Texas through a commercial partnership with Cerebras, Semafor reported. Cerebras was significantly dependent on G42 for revenue, which accounted for 87 percent of its revenues in the first half of 2024.
Despite these U.S. partnerships, G42’s historical ties to China have been a point of concern for the U.S. government. G42 previously used Huawei hardware for its data centers, leading U.S. officials to warn the UAE government that it must choose between partnering with U.S./Western firms or Chinese companies in AI, a Congress.gov report noted.
Following U.S.-Emirati negotiations, UAE officials promised to work more closely with U.S. companies on data security, divest from Chinese companies, and G42 specifically agreed to sever ties to Huawei and other firms. Soon after G42 agreed to sever ties with Huawei, Microsoft and OpenAI both launched partnerships with G42.
The Microsoft-G42 deal is governed by an “Intergovernmental Assurance Agreement” developed in close consultation with both the UAE and U.S. governments. Some lawmakers have called for a further U.S. assessment of the risks of U.S. corporate partnerships with G42, the Congress.gov report added. The US had approved the sale of cutting-edge Nvidia chips, including a sizable order of H100 models, to G42 earlier in 2024, Semafor also reported.
