“Why are we placing information facilities and analysis hubs in Dubai?” U.S. Rep. Ro Khanna in a video posted on X in Might. “We must always have these high-paying new know-how jobs in the US. What occurred to ‘America First’?” U.S. President Donald Trump’s czar for synthetic intelligence, David Sacks, fired again: “The deal has an identical funding provision, so UAE will fund the build-out of AI infrastructure within the U.S. a minimum of as massive and highly effective as that in UAE. And in UAE, the overwhelming majority of the compute will likely be owned & operated by American cloud corporations, to serve the area in addition to the World South.”
Theirs was not a mere rhetorical spat. It centered on the consequence of Trump’s Might journey to the Gulf: the U.S.-UAE AI Acceleration Partnership, anchored by an AI information heart in Abu Dhabi. Stargate UAE, an infrastructure platform and the inaugural venture within the OpenAI for International locations initiative, will likely be a part of the info heart, which will likely be one of many world’s largest. Developed in shut coordination with the U.S. authorities and personally championed by Trump, the partnership entails a twin funding: a 1-gigawatt (GW) cluster in Abu Dhabi (with the primary 200 megawatts anticipated to go dwell in 2026) and an Emirati dedication to broaden U.S. Stargate infrastructure. Saudi Arabia is adopting an analogous method by means of HUMAIN, the Public Funding Fund’s flagship AI enterprise. Backed by offers with, amongst others, Nvidia and Qualcomm, HUMAIN can also be designed as a Saudi platform with regional and world ambitions, focusing on mannequin coaching, inference, and AI providers throughout rising markets—all firmly embedded within the U.S. tech stack. Collectively, the UAE and Saudi Arabia, and the broader Gulf area, are positioning themselves as potential backends of AI for rising markets throughout Asia and Africa, laying the groundwork for a U.S.-aligned mannequin of AI partnerships that might, over time, outpace China within the world AI race.
“Why are we placing information facilities and analysis hubs in Dubai?” U.S. Rep. Ro Khanna in a video posted on X in Might. “We must always have these high-paying new know-how jobs in the US. What occurred to ‘America First’?” U.S. President Donald Trump’s czar for synthetic intelligence, David Sacks, fired again: “The deal has an identical funding provision, so UAE will fund the build-out of AI infrastructure within the U.S. a minimum of as massive and highly effective as that in UAE. And in UAE, the overwhelming majority of the compute will likely be owned & operated by American cloud corporations, to serve the area in addition to the World South.”
Theirs was not a mere rhetorical spat. It centered on the consequence of Trump’s Might journey to the Gulf: the U.S.-UAE AI Acceleration Partnership, anchored by an AI information heart in Abu Dhabi. Stargate UAE, an infrastructure platform and the inaugural venture within the OpenAI for International locations initiative, will likely be a part of the info heart, which will likely be one of many world’s largest. Developed in shut coordination with the U.S. authorities and personally championed by Trump, the partnership entails a twin funding: a 1-gigawatt (GW) cluster in Abu Dhabi (with the primary 200 megawatts anticipated to go dwell in 2026) and an Emirati dedication to broaden U.S. Stargate infrastructure. Saudi Arabia is adopting an analogous method by means of HUMAIN, the Public Funding Fund’s flagship AI enterprise. Backed by offers with, amongst others, Nvidia and Qualcomm, HUMAIN can also be designed as a Saudi platform with regional and world ambitions, focusing on mannequin coaching, inference, and AI providers throughout rising markets—all firmly embedded within the U.S. tech stack. Collectively, the UAE and Saudi Arabia, and the broader Gulf area, are positioning themselves as potential backends of AI for rising markets throughout Asia and Africa, laying the groundwork for a U.S.-aligned mannequin of AI partnerships that might, over time, outpace China within the world AI race.
As proven by Khanna and Sacks’ change, although, the deal—coupled with the rescission of the Biden‑period AI diffusion rule—has sparked sharp debate, even earlier than its official rollout. The stakes embrace the query of which superpower will personal the infrastructure for a very powerful know-how of our time.
Critics in what could be known as the “AI nonproliferation camp,” together with Khanna, argue that Washington is successfully the US’ strategic computing energy to the Center East. They argue that compute—the capability to course of information, powering the whole lot from fundamental calculations to superior AI techniques—is the Twenty first-century equal of enriched uranium, a vital lever of nationwide energy that must be distributed cautiously, if in any respect. The US controls roughly 75 p.c of the world’s frontier‑class compute, powered largely by Nvidia GPUs and Arm‑based mostly designs. Many in Washington imagine that China, constrained by export controls, can’t quickly assemble, not to mention deploy, a aggressive AI stack at scale. These within the “nonproliferation” camp argue that profitable the race to synthetic common intelligence requires sustaining this benefit. They assist a tiered world system, through which cutting-edge chips are granted solely to trusted allies, whereas the remainder of the world—together with energy-rich Gulf states—is left on the surface.
Sacks and Sriram Krishnan, a senior White Home advisor on AI, lead what could be known as the “world diffusion of the U.S. AI stack” camp. They argue that the chokepoint in AI infrastructure is not chips however electrical energy. With U.S. information heart demand set to eat as much as 9 p.c of the nationwide grid by 2030, they observe that the US can’t construct excessive‑voltage traces or substations quick sufficient to maintain tempo. Gulf companions, against this, can finance GW‑scale campuses as little as $0.03 per kilowatt-hour and produce them on-line in months. In Sacks’s view, tapping this surplus energy is the one credible solution to scale compute rapidly sufficient to maintain the US in entrance of the worldwide AI race.
This camp additionally contends that the technological hole between the US and China is much narrower than AI nonproliferation advocates recommend. Huawei’s Ascend chips, paired with the DeepSeek mannequin suite, present Beijing with an end-to-end home stack that’s already seeding pilot tasks overseas. Huawei is approaching close to parity with U.S. chips and China’s broader push for technological self-reliance, particularly in sectors like automotive chips—is accelerating, with main EV makers like BYD and Geely getting ready automobiles utilizing 100% home chips by 2026 and mature-node chip capability in China projected to achieve almost 40% of worldwide provide by 2027, up from 31% in 2023. Even when these targets aren’t totally met, the course of journey is obvious, and the shift already carries world implications. This could immediate a critical rethinking of the present state of U.S.-China tech competitors. And since this competitors isn’t taking place in a vacuum, different nations are already going through a selection between two stacks: the U.S. stack or the Chinese language stack. In reality, that selection is unfolding now.
Malaysia’s flip-flopping over a Huawei-backed nationwide AI cloud is a living proof. As Sacks stated, “As I’ve been warning, the total Chinese language stack is right here. We rescinded the Biden Diffusion Rule simply in time. The American AI stack must be unleashed to compete.” As soon as Ascend {hardware} and DeepSeek fashions unfold throughout rising markets, Sacks thinks they may erode the present U.S. compute edge and speed up China’s momentum within the race for AI dominance. Even when Malaysia finally opts out of deploying the Chinese language stack, the stack is already right here and can unfold. Based mostly on China’s increasing digital footprint, a rising variety of international locations might probably undertake its AI stack, together with Bangladesh, Pakistan, Thailand, South Africa, Algeria, and Egypt. The Chinese language stack has no scarcity of potential prospects. It’s solely a matter of time.
For that reason, the U.S.-UAE and the US-Saudi partnerships shouldn’t be seen as offshoring compute capability—that is capability that didn’t exist within the first place. The US’ present compute infrastructure is already strained by extreme home bottlenecks in power and capital, which makes the UAE and Saudi Arabia obligatory companions for increasing U.S. compute capability. In response to the U.S. Division of Vitality (DOE), information heart energy demand in the US is projected to rise sharply, from about 4.4% of complete U.S. electrical energy consumption in 2023 to between 6.7% and 12% by 2028. Assembly this surge—requiring 50 to 60 GW of extra infrastructure for generative AI alone—is constrained by seven-year timelines for brand new energy technology and grid connections, compounded by challenges in securing capital and land permits, in addition to rising group resistance. On this context, the U.S.-UAE partnership and the U.S.-Saudi offers—and the UAE and Saudi Arabia’s dedication to financing the buildout of U.S. compute infrastructure—characterize essentially the most practical path for Washington to scale up compute capability in partnership with keen and succesful nations.
If the query of compute capability is finally considered one of power—and if the U.S.-UAE partnership helps resolve that constraint—some critics nonetheless argue that such cooperation must be restricted to nations that share the US’ democratic values. This argument rests on the belief that compute is just too delicate to share. In actuality, nevertheless, compute just isn’t a categorized asset; it’s a commodity. Like oil within the twentieth century, the worth lies not in secrecy however in capability, scale, and dependable entry. You don’t must reveal the supply code of an iPhone to promote it; the worth lies in mass manufacturing, efficiency, and distribution. Compute works the identical method. What issues is who can construct and scale the infrastructure, not who hides it behind a firewall.
By means of its partnership with the US, the UAE, Saudi Arabia, and the broader Gulf might place itself as a backend compute service supplier for international locations, notably throughout rising markets, that lack the capability to coach and run AI techniques however are desirous to leverage them for nationwide targets. In some ways, that’s the essence of sovereign AI.
If the US turns away from keen companions just like the UAE and Saudi Arabia—nations which have demonstrated a transparent desire for the U.S. stack by working with U.S. corporations and investing in U.S. AI infrastructure—out of suspicion or by means of the slender lens of the democracy-versus-autocracy framework, Washington dangers making a vacuum that China will likely be fast to fill. In contrast to in conventional areas of geopolitical competitors, the place Beijing has but to show that it might probably exchange the US as a worldwide safety supplier, know-how is a special story. On this space, China has outpaced Western companies, constructing a worldwide community that provides its stack with no strings connected. Chinese language 5G {hardware}, renewable applied sciences, electrical automobiles, and different mass-produced applied sciences have already received the worldwide race. The selection, then, just isn’t between trusting autocracies or preserving democratic values; it’s between main the worldwide diffusion of U.S. AI infrastructure or standing idle because the Chinese language stack turns into the default.
