1. The Moment of Forced Choice
April 15, 2024. Abu Dhabi. Microsoft President Brad Smith arrived at G42 headquarters. The stated purpose was a "partnership announcement." The reality was different. Across the table sat G42 CEO Peng Xiao — a technology executive of Chinese heritage who had surrendered his American citizenship and obtained UAE nationality. The documents he was about to sign would fix all his coordinates in one direction. Microsoft had agreed to invest $1.5 billion — roughly ₩2 trillion — in G42. Smith joined the G42 board. G42 committed to migrating its entire AI development and deployment stack to Microsoft Azure.
The real terms of the deal lay elsewhere. Four months earlier, in December 2023, G42 had formally severed ties with Huawei. Two months after that, in February 2024, it sold its entire stake in ByteDance and other Chinese firms. The decision had come from G42 chairman Sheikh Tahnoon bin Zayed Al Nahyan — a central figure in the UAE royal family and the national security adviser. In exchange for cutting China loose, G42 gained access to American technology. It had been more than a year since the U.S. House Select Committee on the Chinese Communist Party named G42 as a conduit into Chinese military networks and pressed for scrutiny.
Axios reported the deal as part of Washington's effort "to align a key Middle Eastern company to the U.S. side while blocking China's access to AI technology." This was not a venture investment. It was an official declaration of geopolitical alignment.
That is the core of what this chapter argues. The UAE had tried to hold the "middle" between the United States and China for as long as possible. When the technology-hegemony competition intensified, that middle space contracted. It was the moment when the limits of what capital can buy became visible.
And at the frontier of those limits stood Peng Xiao.
2. "Buy the Future Before the Oil Runs Out"
In Chapter 8 we saw Indonesia build leverage from resources. Indonesia locked down nickel — a subterranean resource — with an export ban and pulled its position up the value chain. The UAE started from the same resource — oil — but chose a different direction. Where Indonesia used its resource as a weapon, the UAE converted it into capital before the resource ran dry.
The numbers tell the story. Abu Dhabi alone manages sovereign assets totaling $1.7 trillion as of 2024. ADIA (Abu Dhabi Investment Authority) leads with roughly $1 trillion, ranking it among the world's third or fourth largest sovereign wealth funds. Mubadala, the technology-focused fund, holds $330 billion. MGX, an AI-dedicated platform established in 2024, targets up to $100 billion.
Behind this capital accumulation lies a single piece of hard-nosed realism. Oil demand will peak sometime in the 2030s and then decline — and the UAE has internalized that scenario. Rather than waiting for the "post-oil" era to arrive, it is converting peak-revenue years into future assets. The country built on sand above oil is now building its future on sand below AI. The world's first Ministry of AI, established in 2017, and the target that "AI will contribute 40 percent of GDP by 2031" were the official declaration of this transition.
Mubadala shows what the strategy looks like in practice. The fund is the largest shareholder in GlobalFoundries, the world's third-largest semiconductor foundry — a seed planted fifteen years ago when it acquired AMD's manufacturing division, now grown into a strategic position in the semiconductor supply chain. In 2025 alone it poured $12.9 billion into AI and digital infrastructure. MGX has entered as an anchor investor in both OpenAI and Elon Musk's xAI simultaneously. With BlackRock and Microsoft it formed the Global AI Infrastructure Investment Partnership (GAIIP), a $30 billion vehicle for global AI-infrastructure investment.
What the UAE is buying is not yield. It is a stake in the future technology order.
The strategy rests on an energy logic as well. AI data centers consume enormous power. The four Barakah nuclear reactors in Abu Dhabi supply 25 percent of the UAE's total electricity. The Al Dhafra solar plant — one of the largest single solar installations in the world — produces two gigawatts. The share of renewables reached 35 percent in 2024, up from three percent in 2019: a twelvefold increase. Four million solar panels laid across the desert floor underpin the power base for AI compute. Wealth accumulated from oil funds the nuclear plants; the desert itself is tiled with solar panels. The conversion of resources is proceeding on multiple layers at once.
3. The Structural Power of Sovereign Capital
In Chapter 3 we saw the indispensability Lee Kuan Yew deliberately designed. A small country's strategy of honing institutions, recruiting talent, and making itself an entity no great power could swallow. The UAE is also designing. But the method is different. Lee Kuan Yew built indispensability through institutions. The UAE buys indispensability with capital.
There are structural reasons this strategy works.
The starting point is the absence of any need for political consensus. The UAE is a monarchy. The appointment of an AI minister, the establishment of MGX, the pursuit of Stargate UAE — these decisions are executed on a timescale of months, with no parliamentary debate. Compare that with the 3.5 years it took Korea's AI Basic Act (인공지능 기본법) to pass the National Assembly. From the moment G42 cut ties with Huawei to the moment it signed with Microsoft: four months. The speed of strategic reorientation is itself a competitive advantage.
To that is added the possibility of long-horizon investment. A national pension fund carries the liability of retirement payouts, which prevents concentrated bets on high-risk assets. A principal loss becomes a political event. ADIA and Mubadala face no such constraint. They can take positions with a ten- or twenty-year horizon and absorb losses along the way. The case of GlobalFoundries — created in 2009 when Mubadala spun off AMD's manufacturing division and became its largest shareholder — is the proof. Fifteen years of investment are now bearing fruit as a strategic position in the semiconductor supply chain.
When these two conditions combine, the decisive variable activates: capital large enough to access the deepest core of the technology ecosystem. A country of ten million people — 1.33 million of them citizens — holds strategic investment relationships simultaneously with OpenAI, xAI, Microsoft, and BlackRock. When capital is large enough, even a geopolitically small state can become a strategic actor. This is the essential logic of the Type-C archetype.
Stargate UAE, launched in May 2025, is the culmination of this strategy. On a ten-square-mile site (roughly 26 square kilometers) in Abu Dhabi, OpenAI, Oracle, NVIDIA, and SoftBank are jointly building an AI campus. Total target capacity: five gigawatts — the largest outside the United States.
The first 200-megawatt cluster is scheduled for completion in the third quarter of 2026, with an investment of approximately $20 billion. In November 2025, Microsoft announced it would quadruple its AI data-center capacity in the UAE and invest a total of $15.2 billion through 2030.
What the UAE provides is land, power, and capital. The technology stack is held by American Big Tech, but the infrastructure on which that technology stack physically operates is supplied by the UAE. This is the form of indispensability the UAE has designed — not the source of technology, but the necessary condition for technology to function.
A city built on sand depends not on the sand but on the bedrock beneath it. What the UAE is raising on that sand is not only data centers and GPU clusters — it is laying the bedrock of geopolitical trust with capital. How solid that bedrock is remains the central question of this pivot.
4. The AI a Country of Ten Million Built
The most striking example of the sovereign-capital strategy is TII (Technology Innovation Institute) and the Falcon LLM.
TII is an applied-research institute established under the Abu Dhabi Advanced Technology Research Council. The majority of its research staff are foreign scientists recruited from more than 50 countries worldwide. Falcon 180B, unveiled by the institute in September 2023, recorded the highest performance of any open-source LLM at the time of its release. 180 billion parameters. 3.5 trillion tokens of training data. Seven million GPU-hours consumed. On the Hugging Face leaderboard it scored 68.74 — surpassing Meta's Llama 2 and exceeding OpenAI's GPT-3.5.
The numbers matter less than the meaning. A national LLM is not a function of population size. The key is concentration of capital and recruitment of global talent. A country of 1.33 million citizens built a language model of global standing. Releasing that model as open source was not a monetization strategy; it was a demonstration of national capability. In 2024 the UAE established the Falcon Foundation and endowed it with $300 million. This is how the UAE approaches AI — not as a commercial service but as a national strategic asset.
The talent-attraction strategy operates on the same logic. The UAE Golden Visa guarantees a ten-year residency. In 2024 alone, more than 12,000 engineers and technology professionals received Golden Visas. The renewal rate is 68 percent. AI specialists, climate-tech professionals, and cloud-computing experts were designated as priority categories for 2025. Together Dubai and Abu Dhabi host more than 1,200 AI companies.
G42 CEO Peng Xiao is of Chinese-American background. He served ten years at MicroStrategy as CTO and CIO. At some point he surrendered his American passport and applied for UAE nationality. He changed countries.
That choice is the compressed version of this entire chapter. Peng Xiao was a man standing at the border of two great powers — Chinese heritage, American education and career, and a third coordinate: the United Arab Emirates. When G42 partnered with Huawei and collaborated with the Chinese technology ecosystem, he was the executor of that strategy. When the U.S. Congress designated G42 "a conduit into Chinese military networks" and called for an investigation, he found himself under the spotlight of the New York Times. In the summer of 2023, word came that the U.S. Commerce Department was tightening surveillance of NVIDIA chip exports transiting the UAE. Closed-door meetings with Chairman Tahnoon became more frequent. The options before Peng Xiao were two — keep the Huawei servers and lose access to American technology, or liquidate the Chinese partnerships and join forces with Microsoft. There was no middle path. Within four months he had removed Huawei equipment from G42 data centers, sold the ByteDance stake, and signed the $1.5 billion contract with Microsoft.
How a single human being carried the weight of that decision is not something we can know. What he said afterward was this: "When I came to this country nine years ago, I was responding to the calling of this nation." The calling of the nation — that calling now meant alignment with the United States. A CEO of Chinese heritage executing the decision to sever ties with China. If capital cannot buy a technologist, design the conditions that make that technologist a citizen — the UAE built that design, and Peng Xiao executed the hardest choice within it.
But Falcon has its limits. Falcon 180B, which was the world's leading open-source LLM at its 2023 release, was overtaken in performance by Meta's Llama 3 and Google's Gemini within less than a year. Competition in AI models demands continuous R&D investment and density of talent. Capital can buy the first leap. Maintaining the frontier is a different problem.
This is the tension that runs through the whole of this chapter. Building a structure on sand and having that structure generate new structures on its own are two different things. For the UAE to become a genuine AI power, the foreign scientists at TII must choose to make the UAE their permanent research base — and that choice is made not by capital but by environment.
5. What Capital Cannot Buy
This strategy has its shadows.
What the G42 episode revealed is the structural limit of the capital-pivot strategy. The UAE had held the "middle" between the United States and China for a long time. Huawei, BGI Group, ByteDance, Sinopharm — G42 maintained simultaneous relationships with the core actors of the Chinese technology ecosystem. The U.S. Congress defined this as "a channel through which advanced American technology is rerouted to China" and applied pressure, and the UAE was ultimately forced to choose. However large the capital, it could not sustain a middle position indefinitely against the logic of great-power technology blockades.
After the Microsoft investment, G42 became dependent on the Azure platform. Stargate UAE has OpenAI, Oracle, and NVIDIA holding the core technology stack. What the UAE provides is land, power, and money; the operational know-how of the technology resides with American Big Tech. The strategy of "buying" indispensability paradoxically deepens technological dependence. This is the inherent contradiction of the Type-C archetype.
The absence of an endogenous technology ecosystem is also a fundamental constraint. TII's researchers are foreign nationals. G42's core technology comes from Microsoft and NVIDIA. The Falcon model was developed by foreign scientists. Purchasing and innovating are different things. Capital can buy advanced technology, but the culture and density of talent that continuously advances that technology cannot be accumulated in a short period. A Golden Visa renewal rate of 68 percent also means 32 percent leave.
There is a demographic vulnerability as well. Of the UAE's total population of 11.35 million, citizens account for just 11.5 percent. The foreign-resident share ranks third in the world. In the national decision-making structure, the 88.5 percent who are foreign nationals have no political rights. If economic conditions deteriorate, they can leave. The question of how robust the human foundation of the technology ecosystem truly is remains open.
Geopolitical risk is a permanent variable. Iran sits across the Persian Gulf. In 2022, Yemen's Houthi rebels struck UAE territory with drones and missiles. What the UAE has erected as an AI hub stands on the unstable ground of Middle Eastern regional insecurity — quite literally on sand. The 26-square-kilometer Stargate UAE campus rises from concrete and steel in the middle of the desert. One of the most concentrated AI infrastructure installations in the world could simultaneously become one of the most exposed geopolitical targets. This is the second meaning carried by the chapter title: a pivot on sand.
6. The Paradox of Type-C
In Chapter 6 we saw Singapore's GIC and Temasek dual structure. GIC handles long-term asset preservation and the acquisition of stakes in global AI frontier companies; Temasek takes the lead in strategic industries and digital-economy dominance across Southeast Asia. Singapore's sovereign-wealth-fund strategy is a combination of capital and institutions — capital deployed on a foundation of rule of law, regulatory predictability, and institutional trust.
The UAE's strategy is different. It seeks to manufacture indispensability through the sheer scale of capital rather than institutional trust. This is the essential wager of the Type-C archetype. How valid that wager is compresses into a single question: can indispensability be bought with capital?
Partially, yes. Stargate UAE is real. OpenAI and NVIDIA are in Abu Dhabi. At the May 2025 summit between President Trump and Mohammed bin Zayed (MBZ), the UAE was granted permission to purchase up to 500,000 NVIDIA processors annually. That came as a package with a pledge of $1.4 trillion in investment into the United States over ten years. In exchange for becoming an overseas anchor of the American AI ecosystem, the United States recognized the UAE as a trusted partner in the AI supply chain. Capital coordinated geopolitical status.
The price of that coordination was severance from China. Plant one flag on the sand and you must lower the other. The "middle" could not be held indefinitely.
But this status is maintained by the scale of capital. The moment capital stops flowing, or the moment a new actor with more capital appears, that status becomes subject to renegotiation. If Singapore's institutional indispensability rests on the structure of rule of law and regulatory predictability, the UAE's capital indispensability requires continuous input. It is a leasing of the in-between space with capital.
Applying the core formula, the UAE's trajectory looks like this. Through an alternative path to technological innovation — oil revenues — concentration of capital was achieved. Deploying that capital into AI infrastructure is a form of institutional redesign that circumvents social instability: more precisely, an attempt to buy time with capital in the absence of institutions. Whether this path is sustainable is a race between how long oil revenues hold and how quickly the strategic conversion of capital translates into endogenous capability.
Technological innovation → Concentration of capital → Social instability → Institutional redesign
Capital can buy time. But capital cannot replace time.
7. What Was Built on Sand
In Book 3 we saw BlackRock's Aladdin system demonstrate a new paradigm — allocating global capital by algorithm. The UAE's sovereign wealth fund strategy is the state-level version of that. It is a combination of the monarchy's strategic judgment and algorithmic portfolio management. The joint formation by MGX and ADIA with BlackRock, Microsoft, and GIP of the $30 billion GAIIP is the physical evidence of that intersection. BlackRock, which runs Aladdin, and Abu Dhabi, which sees AI as "the new oil," joined hands. From the algorithmization of private capital to the algorithmization of state capital — at the intersection of these two actors, a new order of capital allocation is taking shape.
The UAE's strategy is undeniably bold. A nation of 1.33 million citizens has attracted the world's largest AI campus and secured a core investor position in the world's leading AI companies. Yet as the G42 episode showed, this position was not the product of free choice — it was the product of forced choice. Indispensability is not something one is selected for. It is something one creates for oneself.
Peng Xiao's choice compresses that. A CEO of Chinese heritage executed the decision to sever ties with China. If Chairman Tahnoon was the architect of that decision, Peng Xiao was its face. How a country built on sand pivots direction when confronted with a forced choice — that entire process is inscribed in one person's career.
The deeper question the UAE's strategy poses is this: can capital buy time? Building Falcon LLM, attracting Stargate UAE, investing in OpenAI — all of it is an attempt to purchase the time needed for endogenous technology capability to grow. Can the UAE, within that window, convert itself into an ecosystem that generates original technology on its own? There is no answer yet.
How long the pivot the UAE has built on sand will hold depends less on the scale of that capital and more on how deeply that capital converts into endogenous capability. Oil comes from the ground. A technology ecosystem comes from people. People can be recruited with capital, but capital cannot make them put down roots. Capital is a necessary condition for indispensability. It is not a sufficient one.
Korea in the Mirror
The face of Korea that the UAE reflects is a paradox. Korea already possesses what the UAE has been spending decades trying to buy with capital. Yet it cannot deploy that possession strategically.
Scale without strategy. Korea's National Pension Service (NPS) manages assets of roughly ₩840–900 trillion as of 2025 — approximately $640–680 billion — the world's third-largest pension fund. That is about 60–70 percent of ADIA's scale. But NPS is not strategic capital. Because its source is citizens' retirement premiums, it cannot take the concentrated bets on specific technology sectors that constitute a "national wager." When the UAE can direct $100 billion into AI through MGX alone, Korea has no comparable pool of national strategic capital it can mobilize.
KIC's scale and latitude. The Korea Investment Corporation (KIC) reached a record $232 billion in assets under management as of 2025 — smaller than Mubadala ($326 billion) and less than a quarter of GIC (roughly $770 billion). Unlike the GIC-plus-Temasek dual structure we saw in Chapter 6, KIC operates under investment guidelines set by the Ministry of Economy and Finance and the Bank of Korea. In December 2025, a proposal to reshape KIC into a Temasek model was formally raised. The discussion has begun — but compared with the time it took the UAE to establish MGX and invest in OpenAI, the pace is on a different scale entirely.
The technology core exists — and yet. This is the heart of the paradox. Korea already possesses what the UAE has been trying to buy with capital for decades — semiconductors, AI infrastructure, manufacturing technology. Where the UAE is trying to enter the semiconductor supply chain through a GlobalFoundries stake, Korea holds Samsung Electronics and SK hynix as primary assets. The problem is that no structure connects this technology core to strategic capital deployment. A country with the technology cannot deploy capital strategically. A country with the capital is trying to buy the technology.
A precedent for the US-China forced choice. G42's forced choice is a scenario Korea must study carefully. The UAE held the "middle" with capital and ultimately chose the United States. Korea is holding the "middle" with technology. Samsung and SK hynix's semiconductors are indispensable to the American AI ecosystem and at the same time deeply connected to the Chinese market. What the UAE's experience shows is that this middle position is not permanent. As US-China technology blockades intensify, Korea too may face a "forced choice" moment similar to what G42 experienced.
Where Korea differs. The UAE had narrow options in its capital pivot. Without technology or population, capital was the only card it had. Korea holds many cards — semiconductors, AI manufacturing infrastructure, human capital, and a sovereign wealth fund of some scale. When the UAE is trying to enter the semiconductor supply chain through a GlobalFoundries stake, Korea already holds Samsung Electronics and SK hynix as primary sources. What capital cannot buy, Korea already has. And yet it cannot deploy those assets strategically.
What Korea should learn. The UAE's mirror reflects not the scale of capital but the necessity of strategic concentration. Just as G42 pivoted direction in four months when confronted with a forced choice, Korea too — as US-China technology blockades deepen — may face a moment of forced choice in its semiconductor and battery supply chains. Before that moment arrives, redesigning KIC as strategic capital and building the structure that connects the technology core to capital deployment — that is the lesson the UAE reflects back onto Korea, in reverse. A country that cannot act because it lacks capital and a country that does not act because it has capital but lacks strategy may end up in the same place.
The next mirror is the country that shows Korea its nearest future — Japan, which walked the same road thirty years ahead.