Publications

At the end of 2025, the United States approved, for the first time, the export of the most advanced AI chips produced by Nvidia to Saudi Arabia and the United Arab Emirates, in an initial quantity of tens of thousands of chips for each country. This represents a technological and economic development of historic proportions: Over the coming years, the two states plan to acquire hundreds of thousands of such chips to operate computing and data centers—volumes possessed only by the United States and China—with the explicit aim of holding a significant share of global AI production and storage capacity. Qatar, the third-richest state in the Gulf, is also accelerating its AI capabilities to position itself as a global player in the field.

The three Gulf states identified the rise of AI toward the end of the previous decade and adopted it as the backbone of their economic diversification strategies—“Vision 2030” in Saudi Arabia and Qatar and “Vision 2031” in the UAE. They are exploiting the current “golden window” of high oil and gas prices to finance an accelerated technological leap ahead of the expected decline in global demand for these resources. As Saudi Finance Minister Mohammed al-Jadaan succinctly put it: “Instead of exporting oil—we will export data.”

Riyadh, Abu Dhabi, and Doha view AI as a long-term strategic alternative to oil and gas across several dimensions. First, revenues from energy exports allow them to sustain the “rentier state” model—a state that does not levy taxes, provides welfare and subsidies, and maintains an inflated public sector—in return for legitimacy without political participation by citizens. Gulf rulers present AI as a means of preserving these welfare and subsidy mechanisms through high-paying jobs and alternative state revenues. At the international level, AI resembles oil and gas in being a horizontal sector that affects multiple domains, such as finance, industry, and defense; control over its bottleneck would therefore grant the Gulf states broad and sustained leverage over the global economy.

There is also a security dimension. For decades, the national security of the Gulf states rested on an informal equation: The Gulf provided stable energy supplies, and the United States provided security—as manifested in the 1991 Gulf War. Over the past decade, however, global dependence on Gulf oil has weakened, and accordingly so has US commitment, highlighted starkly by the absence of a response to the Iranian attack on oil facilities in eastern Saudi Arabia in 2019. This event made clear to Gulf rulers that their oil wells were no longer assets of decisive global importance. Consequently, the three Gulf states are cultivating a new strategic asset: They are building a network of massive computing centers—the physical infrastructure in which AI language models used in every household (such as ChatGPT or Gemini) are developed and stored—in partnership with major US technology companies, including Google, OpenAI, Microsoft, Amazon, xAI, Oracle, and Cisco. The rationale underlying this strategy is simple: Just as the global economy once could not afford disruptions in energy supply from the Gulf, so in the future it will not be able to afford disruptions in the operation of these computing centers, and therefore Washington will stand by the Gulf states’ security needs.

In addition, AI serves Gulf rulers’ efforts to deflect criticism over human rights violations and to ignore demands for changes in their conduct, as they have successfully done until now through oil and gas power. By building computing infrastructure that Western firms come to depend on, and by developing local language models, these three states will be able to filter critical or sensitive content and promote their preferred narratives and “soft power” within the global information flow.

Abundant Advantages but Also Difficulties

The ability of the Gulf states to become AI powers stems from an unusual combination of built-in advantages, primarily their cash availability, as the three ruling families control sovereign wealth of roughly $3.5 trillion. This capital is critical in the AI race, which is characterized by extreme “economies of scale”—only actors capable of absorbing enormous research and infrastructure costs can pose meaningful global competition. Second, as major producers of oil and natural gas, the three Gulf states have abundant, inexpensive electricity—AI computing processes are voracious consumers of power (a single chip consumes electricity comparable to that of a suburban household). While much of the world is rapidly heading toward electricity shortages insufficient to meet surging AI demand, the power reserves of the Gulf states exceed domestic consumption.

Added to this is an abundance of available land in the vast desert—long-term data and model storage requires enormous physical space—and, of course, efficient centralized governments with regulatory flexibility and without bureaucracy or other democratic constraints. The geographic location of the states, along the main route of submarine fiber-optic cables connecting Europe, Asia, and Africa through which most global data flows, is also ideal for computing centers, as it allows minimal latency. The combination of all these advantages enables the Gulf states to build computing and data storage centers faster and at larger scales than in the United States or Europe, and at lower operating costs for the companies that use them.

At the same time, the path of the three states to the top of the AI world is not without challenges. First, they suffer from an acute shortage of the required human capital, forcing them to import experts from abroad, which leads to rapid adoption of technologies without a solid scientific base and exposes the industry to the risk of abandonment in times of crisis. Second, there is decisive dependence on imported foreign hardware. Finally, their desert climate, where summer temperatures reach around fifty degrees Celsius, requires intensive cooling of computing centers. This process consumes increasing quantities of water (their water consumption relies on desalination, which itself requires energy) and creates a threatening cycle that could increase carbon emissions and warming, and even disrupt energy and water supplies to the population.

Moreover, given the authoritarian nature of the Gulf states, their ethical regulations for privacy protection and dual-use (military and civilian) applications are far weaker than in democracies. The lack of separation between AI research and development and the security and intelligence mechanisms is particularly pronounced in the UAE: Sheikh Tahnoun bin Zayed, the influential brother of the UAE’s president, oversees both. Indeed, Emirati use of civilian-cover technology for intelligence purposes was already exposed in the past when a popular Emirati messaging app was revealed to be a government espionage tool. Consequently, Western concern is growing that models and data could be exploited for internal repression, potentially deterring companies from storing sensitive models and “weights” (training outputs) in the computing centers being built in the Gulf.^[1]

Situation Assessment

The UAE, which established a dedicated government ministry for AI in 2017, is the pioneer and leader among the three Gulf states. Its AI industry is built as an ecosystem centered in the capital, Abu Dhabi, and includes the semi-governmental technology company G42, the governmental Technology Innovation Institute (TII), and the sovereign AI investment fund MGX. To date, the UAE has established 35 computing centers within its territory, the flagship of which is “Stargate”: a hyperscale computing center expected to open gradually from 2026 and serve US technology giants with computing capacity reaching five gigawatts—the largest outside the United States. South Korea will assist in supplying electricity to the center (whose consumption is equivalent to that of New York City), including through nuclear reactors it previously built for the UAE.

Saudi Arabia joined the AI race somewhat later than the UAE, but it is investing larger sums and may overtake it within a few years, having set for itself the ambitious goal of becoming the world’s third-largest AI player after the United States and China. The Saudi industry is centrally managed by HUMAIN, a national AI company that functions as a closed system directly subordinate to the Saudi Crown Prince and de facto ruler, Mohammed bin Salman. Essentially, Riyadh wants HUMAIN to replicate within the AI sector the successful model of the national oil company Aramco, which controls the entire lifecycle of Saudi oil, end-to-end, thereby enabling the kingdom to maximize its influence on the global economy by dictating supply and prices. HUMAIN currently operates twenty computing centers across the kingdom and plans to reach computing capacity exceeding six gigawatts over the next decade.

Qatar’s AI industry is advancing at a far more modest pace than those of Saudi Arabia and the UAE, and its goals are less clearly defined. To date, Qatar has focused primarily on integrating AI applications of international companies into government services and digitization, and on building a regional cloud infrastructure “hub”—mostly through the semi-governmental telecommunications company Ooredoo. Unlike its neighbors, which are investing massively in building computing centers, Qatar currently operates only five such centers, expected to reach at most about half a gigawatt by the end of the decade. Last year, Qatar established a national AI company, QAI, to manage the field under a single umbrella and invested in the US AI startup Anthropic (which operates the Claude language model). The combination of knowledge accumulated through international partnerships, cloud infrastructure for data storage, and exceptional electricity availability in a state that holds over a tenth of global gas reserves is intended to position the tiny emirate as a significant global AI player, even if not on the scale of its larger Gulf neighbors.

At the core of the ambition of the three Gulf states in AI lies the desire to achieve “AI sovereignty”—that is, control over the entire value chain: capital, expertise, processing and development of models, and storage of models and data—in a manner that keeps accumulated knowledge and outputs in their hands and leaves as little dependence as possible on foreign states and companies.^[2] Accordingly, unlike the common Gulf pattern of purchasing “off-the-shelf” products and services in their entirety, the three states generate their own data and store it in sovereign local clouds. The Saudis and Emiratis have even developed sovereign open-source language models, ensuring that foreign suppliers will not control the code or its operation. The UAE’s Minister of Economy, Abdulla bin Touq Al Marri, clarified this point, declaring in 2025 that “AI is a state of sovereignty. Like you spend budgets on defense, on cyber, you have to spend budget on AI. AI is a sovereignty for every single nation.”

The gap between vision and reality lies in human capital. For this reason, the Gulf states operate local training programs for engineers, data scientists, developers, researchers, and other professions essential to the AI industry. In Saudi Arabia and the UAE, these programs are massive in scale, and the Saudis have declared their intention to train 100,000 experts in these fields by 2030. In both states, training the future generation has already begun in school where courses on AI skills were introduced this year and continues in local academic institutions. In Abu Dhabi, the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) was established in 2020, the world’s first academic institution dedicated to the field, while in Saudi Arabia, universities for science and technology operate in Jeddah (KAUST) and Riyadh (KACST). Qatar, by contrast, runs a more modest AI training program at Qatar University and Hamad bin Khalifa University. Institutions in all three states recruit leading global faculty and cooperate with US companies, including Microsoft, Nvidia, and Google, in practical training.

In addition, in Saudi Arabia, the government also operates “AI Zones” (training and AI development environments in partnership with Amazon), AI technology incubators (providing mentoring and infrastructure to local startups), and the “Quantum Valley” (an ecosystem for quantum computing research—a technology expected to drive AI in the future).

Together, the Gulf states' training initiatives generate a growing pool of skilled talent for local AI industries while keeping accumulated knowledge, data, and innovations within the three states and under government control. Their aspirational ambition is not merely to catch up with the West and China but to develop the future technology themselves.

The Great-Power Struggle

The intensive cultivation of local knowledge does not change the fact that, in the foreseeable future, Gulf AI industries depend on imported hardware produced almost exclusively in the United States and China. The United States currently enjoys a clear technological advantage: Two US companies, Nvidia and AMD, are the only ones in the world capable of producing the advanced AI chips and GPUs required to train large models. During former President Joe Biden’s term, the US administration adopted a “containment” approach: restricting chip exports to “middle states” not fully aligned with the United States—such as the Gulf states—out of concern over the leakage of US technology to China. By contrast, the current administration under Donald Trump, encouraged by the technology industry, supports a rapid “diffusion” approach of US chips to middle states to prevent Chinese companies from filling the gap. Beyond economic interests, this is part of the superpower competition over global electricity capacity, which, in the coming years, is expected to be insufficient for surging AI use—the United States wants abundant Gulf electricity to serve its hardware and companies, not China’s.

Proponents of the diffusion approach argue that, over time, it will increase the purchasing states’ dependence on the United States, given US technological superiority. Indeed, at least in the UAE’s case, it has been willing to distance itself somewhat from China in order to gain access to US technology. In 2022, the United States discovered that technology transferred by the Emirati company G42 to the Chinese telecommunications firm Huawei—linked to Beijing’s intelligence service and military—was used to help China upgrade missile capabilities that threaten US aircraft. Under pressure from Washington, the Emiratis agreed to sever the ties of G42 with Huawei to maintain access to advanced US systems. Emirati officials later claimed that Abu Dhabi then made a strategic decision to “go all-in” on US technology to realize its AI ambitions. Similarly, in 2025, as a condition for approval of the sale of Nvidia’s most advanced Blackwell chips, Saudi Arabia and the UAE were required to physically and systematically separate Blackwell-based infrastructure from any systems using Huawei technology.^[3]

At the same time, although the three Gulf states lean toward US technology, they continue to hedge between the two global powers in AI and in other fields. First, access to US chips is always at risk of policy change or reprioritization in Washington under future administrations. Second, keeping pace with the rapid expansion of their AI industries requires them to continue purchasing Chinese equipment and technological services, as those are tens of percent cheaper than US equivalents and typically free of regulatory conditions related to human rights or dual-use applications. As a result, Chinese technology companies, including Huawei, Alibaba, and Tencent, supply the three Gulf states with data and cloud infrastructure and cooperate in training and knowledge development, much like US firms do. In fact, Gulf AI industries will be the first globally in which US and Chinese technologies coexist side by side.

Against this backdrop, the Gulf states are becoming a central arena in the Great-Power Competition over the technological order of the 21st century, particularly between the US-led IMEC initiative and China’s “Digital Silk Road” (the technological arm of the Belt and Road Initiative. Both seek to shape the physical and digital connectivity linking Asia, the Middle East, and Europe. The Gulf states leverage their wealth and AI-related advantages to extract concessions from both China and the United States, not only in technology but also in diplomacy. Above all, the three Gulf rulers hope that the commitments they made last year to President Trump to invest over $3 trillion in the US market in the coming years, with an emphasis on AI partnerships, will incentivize future administrations in Washington to continue supplying them with both advanced chips and coveted military protection.

Internal and Regional Implications

The parallel AI drive of the three Gulf states adds a new arena to the ongoing rivalry among them, manifested both in the economic sphere—where all three compete to attract investment, companies, and skilled workers—and in geopolitical arenas such as Sudan and Yemen, which have flared up recently. Hosting sensitive computing infrastructure heightens the ingrained need of all three to protect their territory from external threats and regional conflicts.

Moreover, their massive investment in AI will be tested against the possibility of a technological “bubble” in the field—overblown expectations that could burst in a sector still evolving rapidly—leaving costly investments and infrastructure without the returns they seek. It should be noted that AI development in the three monarchies is also embedded into a broader social “experiment” centered on changing living environments while preserving tribal and monarchical order that lacks democracy and equality. Especially in the Saudi case, where change is accompanied by the marginalization of the influential religious establishment, there is potential for internal backlash—certainly in the absence of promised economic returns. In this context, it is worth noting that the Gulf states have experience with overambitious projects that collapsed when confronted with reality (for example, Saudi Arabia’s NEOM mega-city, whose grandiose plan was scaled back in favor of diverting resources to AI), remaining partially implemented or as “white elephants.”

In particular, cultivating a critical mass of a highly educated, technologically skilled population places the three monarchies in unfamiliar territory in the Gulf and even in the Arab world. The traditional patronage model—in which the regime provides welfare and services in exchange for political loyalty—will be challenged by a strata of educated specialists with in-demand expertise connected to the global economy. The expansion of this class among foreign resident populations in the three states—especially in the UAE and Qatar, where citizen populations are small—could alter the nature of state–society relations, insofar as local AI sovereignty will, in practice, depend on their service. The ruling families will need to prevent this shift in the socio-political balance from undermining the existing authoritarian political order, as they did, not without difficulty, during the transition from traditional economies to the oil era in the previous century.

Implications and Recommendations for Israel

The rapid empowerment of the Gulf states in artificial intelligence poses an unprecedented challenge to Israel’s regional technological superiority, which for decades rested on a small but high-quality research system. The Gulf states’ ability to attract global human capital—including from Israel—alongside access to chips and computing infrastructure on scales unmatched in Israel, rapidly narrows their gap with Israel in STEM human capital (science, technology, engineering, and mathematics), thereby eroding Israel’s relative advantage. This trend is expected to divert capital flows from Israel toward Riyadh and Dubai, which are increasingly becoming a competing “Silicon Wadi” in the Middle East.

The risk of losing Israel’s technological edge is heightened by the gap between the long-term national strategies adopted by the three Gulf states in the previous decade and Israel’s prolonged delay in formulating a multi-year national AI strategy. An August 2025 report by the National Committee for Accelerating Artificial Intelligence, known as the Nagel Committee, explicitly warned: “States that fail to organize quickly and efficiently will fall behind irreversibly.”

In the security sphere, sovereign AI infrastructure will enable Gulf states to develop advanced military and intelligence capabilities previously exclusive to Israel in the region. The immediate threat comes from Qatar due to its hostile stance toward Israel, expressed in support for Hamas and hostile propaganda through Al Jazeera and other platforms. Sovereign AI would allow it to operate far more sophisticated influence systems—Israel must prepare for AI-based cognitive warfare emanating from Doha.

There is also broader geopolitical significance for Israel: The growing infrastructural and technological value of the Gulf states will counterbalance Israel’s relative advantages in intelligence and cyber for the United States, further cementing the Gulf states’ existing status as Washington’s preferred strategic partners. This trend may further erode Israel’s historical regional standing as the state through which “the road to Washington” passes and could damage the special relationship between Israel and the United States when interests clash with those of the Gulf states.

In particular, attention should be paid to the fact that in order to meet exponential electricity demand, Saudi Arabia is expected to accelerate development of its nuclear power program (a capability it currently lacks), a move requiring close monitoring and continuous dialogue with the United States regarding risks of diversion to military uses. A parallel effort by the Gulf states is expected in air defense, aimed at protecting computing infrastructure in the event of external attacks. In this context, Israeli expertise in the field is well known in the UAE and Saudi Arabia, presenting an opportunity to expand existing cooperation.

Thus, alongside challenges, the rise of the Gulf states as AI powers also creates opportunities for Israel. Smart integration into the emerging regional technological ecosystem could help advance scientific and economic cooperation, through which normalization could be strengthened. Israel’s high-tech industry largely complements that of Saudi Arabia and the UAE: It lacks bureaucratic and regulatory flexibility, centralized management, physical conditions, and heavy government investment—but excels in what the Gulf states lack: human capital, innovation, initiative, and research capabilities, which can synergistically complement Saudi and Emirati financial and infrastructural power.

At the same time, this equation, which until now formed a solid basis for Israel–Gulf relations, will become increasingly less relevant as the Gulf states advance in accumulating sovereign knowledge. Israel is therefore in a race against the clock to extract maximum benefit from its relative advantage while it still exists.

Accordingly, Israel must formulate a long-term national AI strategy focused on nurturing human capital and reducing brain drain, leverage its current advantages in software, science, and regulation to build cooperation frameworks with the UAE and, where possible in the absence of diplomatic relations, with Saudi Arabia, while at the same time monitoring and prohibiting the transfer of critical Israeli knowledge and technologies to Qatar. Alongside close monitoring of developments with military and intelligence potential and deepening strategic dialogue with the United States, multinational initiatives such as “Pax Silica” could serve as a basis for effective regional regulation.

[1] In this context, OpenAI CEO Sam Altman stated that the UAE serves as a “regulatory sandbox” in which AI regulations can be tested in a controlled manner. This characterization drew criticism for disregarding potential human rights violations that may be associated with such experimentation.

[2] An illustration from Israel of a sudden policy shift of a foreign company can be seen in Microsoft’s decision in September 2025 to suspend some of the cloud and AI services it was providing to the IDF due to international pressure following the war in Gaza. This scenario illustrates why the Gulf states insist on establishing AI sovereignty.

[3] There is no known comparable commitment on the part of Qatar, which also uses Huawei technologies. However, the Nvidia chips that Qatar has deployed thus far are not from the Blackwell generation (supplied to Saudi Arabia and the UAE) but rather from a less advanced generation whose sale the United States has approved for China.