In November 2025, the U.S. Commerce Department approved the sale of 70,000 advanced AI chips to two companies most Americans have never heard of: G42, a state-backed AI firm in Abu Dhabi, and Humain, a subsidiary of Saudi Arabia’s sovereign wealth fundA state-owned investment fund that manages national savings or commodity revenues on behalf of a government, typically for long-term economic benefit.. Each will receive computing power equivalent to 35,000 of Nvidia’s most powerful Blackwell GB300 processors.
The deal is the largest single export authorization of frontier AI hardware to the Middle East in history. It signals something bigger than a transaction: the emergence of the Persian Gulf as a third pole in the global AI race, alongside the United States and China.
What Just Happened
The approval came after months of behind-the-scenes negotiations that began during President Trump’s May 2025 visit to the Gulf. During that trip, the administration formally rescinded the Biden administration’s AI Diffusion Rule, which had classified Gulf nations as “Tier 2” countries subject to chip export limits. Trump replaced that framework with bilateral deals negotiated country by country.
The first major partnership to emerge was between Humain and Nvidia, announced on May 13, 2025. Humain committed to building AI data centers in Saudi Arabia with a projected capacity of up to 500 megawatts, powered by several hundred thousand of Nvidia’s most advanced GPUs over five years. The first phase: an 18,000-unit GB300 Grace Blackwell supercomputer.
“AI, like electricity and internet, is essential infrastructure for every nation,” Nvidia CEO Jensen Huang said at the announcement. “Together with Humain, we are building AI infrastructure for the people and companies of Saudi Arabia.”
On the Emirati side, G42 is building Stargate UAE, a 1-gigawatt AI compute cluster in partnership with OpenAI, Oracle, Cisco, Nvidia, and SoftBank Group. It is part of a larger 5-gigawatt AI infrastructure hub designed to serve compute needs across the wider region.
The Price of Admission: Ditching China
These chips did not come free. Both countries paid a geopolitical price: severing ties with Chinese technology companies.
G42’s journey is instructive. In 2024, Microsoft invested $1.5 billion in the company, but only after G42 agreed to strip Huawei equipment from its systems and divest from Chinese firms including TikTok parent ByteDance. Humain separately pledged not to purchase Huawei equipment.
The deals include strict security protocols to prevent diversion to China. Companies must declare planned uses, storage locations, and security measures against unauthorized transfers. G42’s chips will operate under a “Regulated Technology Environment” approved by the Bureau of Industry and Security. These controls come amid ongoing concerns about sophisticated schemes to smuggle advanced chips to China through transshipmentThe routing of goods through an intermediate country or facility before reaching the final destination, sometimes used to obscure the true end-user. networks.
“Our technology ecosystem today is firmly anchored in partnerships with U.S. and allied companies,” G42’s Talal Al Kaissi told Rest of World. The message was unmistakable: the Gulf chose American silicon over Chinese alternatives.
Why the Gulf, Why Now
The Gulf states bring something the United States desperately needs for AI expansion: capital and energy. As the Council on Foreign Relations noted in its analysis, “What the Gulf countries lack in terms of semiconductor prowess and AI talent, they make up for with abundant capital, energy, and accommodating regulations.”
In the United States, building new data centers is constrained by permitting processes and limited grid capacity. The Gulf has neither problem. Saudi Arabia’s Public Investment Fund, with over $900 billion in assets, has already allocated more than $40 billion to AI ventures. The UAE committed to $1.4 trillion in U.S. investment over the next decade as part of the arrangement.
The economic incentive for the Gulf is equally clear. PwC estimates that AI could contribute more than $135 billion to Saudi Arabia’s economy and roughly $96 billion to the UAE’s by 2030, representing around 12-14% of GDP for both nations. For economies built on oil, this is the diversification bet of a generation—especially as regional tensions continue to impact energy markets and supply chains.
The Concerns
Not everyone is celebrating. The Carnegie Endowment for International Peace warned that while tens of thousands of chips represent “a manageable level of risk,” the broader trajectory is concerning. Analysts noted that a deal allowing Gulf companies to amass more than 400,000 H100-equivalent chips “would put those companies on track to build frontier-scale data centers.” And the UAE’s framework authorizes up to 500,000 Nvidia processors per year.
China hawks in both parties worry that Beijing could access advanced AI chips through Gulf partnerships. Chinese firms had been expanding in the region before the deals: Huawei planned a $400 million cloud investment in Saudi Arabia by 2028, and Alibaba and Tencent had launched cloud regions in both countries.
The CFR framed it as a fundamental strategic pivot: “The most recent era of great power competition, the Cold War, was fundamentally bipolar and the United States leaned heavily on the principle of non-proliferation. We are now playing by a new set of rules where the diffusion of U.S. technology is of paramount importance.”
Whether that diffusion strengthens America’s hand or creates a competitor remains the open question. The chips are shipping. The data centers are under construction. The answer will arrive with them.
On November 19, 2025, the U.S. Commerce Department authorized the export of advanced AI semiconductors equivalent to 70,000 Nvidia Blackwell GB300 processors to two Gulf entities: Abu Dhabi’s G42 and Saudi Arabia’s Humain. The authorization, split evenly at 35,000 chip-equivalents per company, represents the most significant proliferation of frontier AI compute outside the U.S.-China axis to date.
The Architecture of the Deals
The Gulf AI buildout is structured in two distinct but parallel tracks.
Saudi Arabia: Humain and the 500MW Vision
Humain, a full AI value chain subsidiary of the Public Investment Fund (PIF), signed its partnership with Nvidia on May 13, 2025, during Trump’s Gulf tour. The deal’s parameters:
- First phase: 18,000 GB300 Grace Blackwell GPUs with InfiniBand networking, forming an AI supercomputer
- Five-year target: several hundred thousand GPUs powering data centers with projected capacity of up to 500 megawatts
- Deployment of Nvidia Omniverse Cloud for physical AI and digital twinA virtual replica of a physical object, system, or process used for simulation, analysis, and optimization without affecting the real-world counterpart. applications
- Large-scale workforce training for Saudi developers in AI, robotics, and simulation
The PIF, with over $900 billion in assets under management, has already allocated more than $40 billion to AI ventures. AMD has a separate agreement worth billions to work with Humain, making Saudi Arabia a multi-vendor AI infrastructure play.
UAE: G42, Stargate, and the 5GW Campus
G42’s chip authorization feeds into several interlocking projects:
- Stargate UAE: a 1-gigawatt AI compute cluster built with OpenAI, Oracle, Cisco, Nvidia, and SoftBank Group
- The broader UAE-U.S. AI Campus: a 5-gigawatt infrastructure hub for regional compute and low-latency inferencing
- Collaborations with Microsoft, AMD, Qualcomm, and Cerebras
The broader framework authorizes the UAE to purchase up to 500,000 advanced Nvidia processors each year (Blackwell now, Rubin and Feynman in coming generations), contingent on a $1.4 trillion investment commitment in the U.S. over a decade. Commerce Secretary Howard Lutnick confirmed that initially, only data centers managed by approved American operators will be eligible to run these systems. G42 will receive 20% of AI processors bound for the UAE in future allocations. This infrastructure buildout occurs against a backdrop of regional supply chain vulnerabilities that could threaten Gulf AI operations.
The Security Framework
Both deals are built on a quid pro quo: American chips for Chinese divestment.
G42’s pivot is the template. In April 2024, Microsoft invested $1.5 billion in G42, but only after the company agreed to strip Huawei equipment from its systems and divest from Chinese firms, including TikTok parent ByteDance. Humain separately pledged not to purchase Huawei equipment.
The operational security architecture includes:
- A Regulated Technology Environment (RTE), a compliance framework approved by the Bureau of Industry and Security
- Mandatory declarations of planned uses, storage locations, and security measures against unauthorized transfers
- U.S. government authorization required before any re-export or transfer to another entity
- Regular reporting on chip location and security compliance
The Carnegie Endowment identified the strongest enforcement lever as the threat of cutting off future export licenses, since AI chips last only three to five years before replacement. “The U.S. government shouldn’t make a deal unless the receiving country will reliably furnish Washington with the information and ongoing leverage necessary to curtail or terminate chip access if security requirements are violated.”
The Policy Shift: From Non-Proliferation to Diffusion
These deals were made possible by the Trump administration’s rescission of Biden’s AI Diffusion Rule in May 2025. The Biden framework had divided the world into three tiers and would have capped any single Gulf company’s chip access at approximately 320,000 H100-equivalents by 2027, though the total across multiple companies could have been significantly higher. Trump replaced it with a bilateral, deal-by-deal approach.
The Council on Foreign Relations characterized this as a fundamental shift in U.S. strategic logic: “The most recent era of great power competition, the Cold War, was fundamentally bipolar and the United States leaned heavily on the principle of non-proliferation. We are now playing by a new set of rules where the diffusion of U.S. technology — and an effort to box out Chinese technology — is of paramount importance.”
The strategy has a logic: if the Gulf is going to build massive AI infrastructure regardless, better it runs on American chips under American oversight than on Chinese alternatives with no U.S. visibility. Chinese firms had been making inroads: Huawei planned a $400 million Saudi cloud investment by 2028, and Alibaba and Tencent had both launched cloud regions in Saudi Arabia and the UAE. The enforcement challenge is real, as demonstrated by recent indictments involving sophisticated chip transshipmentThe routing of goods through an intermediate country or facility before reaching the final destination, sometimes used to obscure the true end-user. schemes designed to circumvent export controls.
Risk Calculus
Carnegie’s analysis offers the clearest risk framework. At the current authorized level of 35,000 GB300-equivalents per company, the risk is “manageable” for the goal of keeping frontier AI development in the United States. “This number of chips would still be enough for some concerning activities, such as nuclear weapons modeling and sophisticated intelligence analysis, but restricting chips at this level is unlikely to be practical.”
The concern is trajectory. Carnegie warned that “a deal that allows Saudi or Emirati companies to amass more than around 400,000 H100-equivalent chips over the next two years would put those companies on track to build frontier-scale data centers.” The UAE’s 500,000 processors-per-year framework is well above that threshold.
The economic stakes cut both ways. PwC estimates AI could contribute over $135 billion to Saudi Arabia’s economy and roughly $96 billion to the UAE’s by 2030, representing 12-14% of GDP. For the Gulf states, this is their post-oil economic strategy made tangible—a crucial hedge as traditional energy market dynamics face increasing volatility. For Washington, that economic integration creates both leverage and dependency.
The CFR’s bottom line captures the strategic gamble: “The question remains whether the diffusion of the most powerful dual-useGoods, technologies, or knowledge with both legitimate civilian applications and potential military uses, typically subject to export controls and international oversight. technologies of our day will bind foreign users to the United States and what impact it will have on the global balance of power.” The chips are shipping. The data centers are under construction. The answer is no longer theoretical.
