Sovereign wealth fund AI investment hit $66 billion in 2025, a figure that would have been unthinkable five years ago.[s] Government-owned investment funds, built on decades of oil revenue, are now pouring that money into data centers, AI startups, and computing infrastructure. The shift marks one of the largest capital reallocations in modern finance: nations that once measured wealth in barrels are now measuring it in GPU clusters and megawatts of compute.
Sovereign Wealth Fund AI Spending: The Numbers
According to Global SWF, state-owned investors directed $66 billion into artificial intelligence and digitalization over the course of 2025.[s] Gulf funds dominated: Abu Dhabi’s Mubadala invested $12.9 billion in AI and digital projects, Kuwait’s Investment Authority committed $6 billion, and Qatar’s Investment Authority put in $4 billion.[s] The seven major Gulf wealth funds together accounted for 43% of all sovereign capital deployed globally, a record.[s]
These are not passive portfolio bets. Saudi Arabia’s Public Investment Fund, which has expanded past $700 billion as part of the kingdom’s Vision 2030 diversification strategy[s], committed $36.2 billion in deals during 2025, making it the single largest sovereign dealmaker of the year.[s]
What They Are Building
In May 2025, Saudi Crown Prince Mohammed bin Salman launched HUMAIN, a PIF-owned company designed to operate across the entire AI value chain: data centers, cloud infrastructure, AI models, and applications.[s] HUMAIN’s CEO, Tareq Amin, stated the company’s goal plainly: “We want to be the third-largest AI provider in the world, behind the United States and China.”[s]
The company started construction on two campuses comprising 11 data centers, each with 200-megawatt capacity, targeting 1.9 gigawatts by 2030 and 6 gigawatts by 2034.[s] By August 2025, HUMAIN had already sold out capacity at every existing and under-construction data center.[s] Notably, 99% of its customers are international, turning the company into an export engine for compute power in the same way Aramco exports crude.[s]
Qatar followed a similar path. In December 2025, Qai, a subsidiary of the Qatar Investment Authority, formed a $20 billion joint venture with Brookfield Asset Management to build AI infrastructure in Qatar and international markets.[s] QIA’s CEO, Mohammed Saif Al-Sowaidi, called the partnership “testament to QIA’s commitment to delivering both local and global impact.”[s]
Abu Dhabi’s approach operates through MGX, a technology fund affiliated with Mubadala. MGX targets $10 billion in deals annually and has backed OpenAI, Anthropic, and xAI simultaneously, viewing each as serving distinct markets.[s] MGX is also a founding partner of the Stargate Project, a $500 billion joint venture with OpenAI, SoftBank, and Oracle announced in January 2025 to build AI data centers across the United States.[s]
Why Oil Money Is Chasing Compute
The logic connecting oil wealth to sovereign wealth fund AI spending has two components. First, Gulf states face a finite timeline on hydrocarbon dominance. Saudi Arabia’s Vision 2030 explicitly aims to reduce reliance on oil by developing new sectors.[s] Data centers, with their steady revenue from global cloud customers, offer a replacement export commodity: compute cycles sold by the gigawatt-hour instead of oil sold by the barrel.
Second, sovereign wealth fund AI investments serve as geopolitical positioning. As one analysis from the Berkeley Political Review noted, SWFs “challenge the traditional separation between markets and state power” because their capital flows function as a mechanism for governments to exert influence beyond their borders.[s]
By making themselves essential to the AI infrastructure that American and European companies depend on, Gulf states gain diplomatic leverage and economic insurance. Roughly 90% of U.S. venture capital flows into software rather than the capital-intensive hardware that AI requires[s], leaving a funding gap that sovereign investors are uniquely positioned to fill with their long time horizons and tolerance for lower short-term returns.[s]
Beyond the Gulf: Norway and France
The sovereign wealth fund AI trend extends well beyond the Persian Gulf. Norway’s Government Pension Fund Global, the world’s largest at $2 trillion, earned $247 billion in 2025, with its annual report citing Nvidia, Alphabet, and Broadcom as the largest contributors to that return.[s] Rather than building data centers, Norway’s fund has deployed AI internally: since 2025, it has used AI models to screen every new equity investment within 24 hours for links to forced labor, corruption, or fraud.[s] In multiple cases, the fund sold positions before the broader market reacted to the flagged risks, avoiding losses.[s]
France, meanwhile, announced roughly €109 billion in AI-related investment commitments from private and international capital, while France 2030 contributes public AI and R&D funding within that broader ecosystem; the 1.2 million GPU target is best attributed to Sesterce’s French AI infrastructure plan rather than a standalone official government goal.[s] Global spending on sovereign AI systems is projected to surpass $100 billion by 2026.[s]
What Comes Next
The question now is whether these sovereign wealth fund AI bets will deliver returns commensurate with their scale, or whether, as some analysts warn, the sector faces bubble dynamics. Norway’s fund CEO Nicolai Tangen has publicly identified an AI bubble as a major risk scenario. The concentration of state capital in a single technology sector creates exposure that could ripple across national budgets if valuations correct sharply.
For the Gulf states, though, the calculus may differ. Even if individual investments underperform, the infrastructure itself, the data centers, the power grids, the fiber optic networks, represents durable national assets. The kingdom that built HUMAIN would still own gigawatts of compute capacity regardless of what happens to AI startup valuations. In that sense, the pivot is less a speculative bet and more a hedged transition: converting depleting oil reserves into permanent digital infrastructure that the world’s economy will rely on for decades.
Sovereign wealth fund AI investment reached $66 billion in 2025, according to data from Global SWF[s], representing a structural reallocation of state capital that moves far beyond the diversification playbook of previous decades. Where Gulf sovereign investors once rotated from hydrocarbons into real estate, hospitality, and financial assets, they are now building vertically integrated AI supply chains: from silicon procurement and data center construction through cloud platforms and model development to application-layer ventures.
Sovereign Wealth Fund AI Capital Flows: Anatomy of the Shift
The aggregate figures merit disaggregation. Mubadala led sovereign AI deployment at $12.9 billion, followed by Kuwait Investment Authority at $6 billion and Qatar Investment Authority at $4 billion.[s] The seven principal Gulf funds accounted for 43% of all sovereign capital invested globally ($126 billion), a historical maximum.[s] Saudi Arabia’s PIF was the single largest sovereign dealmaker at $36.2 billion committed, though the Electronic Arts acquisition skews that figure.[s]
The investment architecture differs by fund. Abu Dhabi operates through a layered structure: MGX, the dedicated AI vehicle affiliated with Mubadala and G42, targets $10 billion in annual deal flow and has taken positions in OpenAI, Anthropic, and xAI, viewing each as occupying distinct market segments (consumer, enterprise, and robotics respectively).[s] MGX also co-founded the $500 billion Stargate Project alongside OpenAI, SoftBank, and Oracle to build AI data centers on U.S. soil.[s]
Infrastructure Plays: HUMAIN, Qai, and the Hardware Gap
Saudi Arabia’s approach through HUMAIN, launched in May 2025 under PIF ownership[s], prioritizes full-stack control. HUMAIN operates across four layers: infrastructure, cloud, data and models, and applications.[s] CEO Tareq Amin has stated the ambition explicitly: “We want to be the third-largest AI provider in the world, behind the United States and China.”[s]
The company’s construction pipeline includes 11 data centers across two campuses, each rated at 200 megawatts, with a buildout cadence of 50 megawatts per quarter through 2026.[s] Capacity targets: 1.9 gigawatts by 2030, 6 gigawatts by 2034.[s] By August 2025, all existing and under-construction capacity was sold out, with 99% of customers located outside Saudi Arabia.[s] The commercial model mirrors petroleum exports: Saudi Arabia extracts value from a physical resource (cheap energy, abundant land) by converting it into a globally traded commodity (compute).
Qatar’s Qai, created under QIA in late 2025, formed a $20 billion joint venture with Brookfield Asset Management focused on AI infrastructure in Qatar and select international markets.[s] Brookfield plans to channel this through its Artificial Intelligence Infrastructure Fund (BAIIF), which aims to mobilize $100 billion in total global AI infrastructure investment.[s]
These infrastructure commitments address a structural gap in private capital markets. Roughly 90% of U.S. venture capital flows into software, not the capital-intensive hardware and physical infrastructure that AI training and inference require.[s] Sovereign funds, with their multi-decade time horizons and tolerance for illiquid assets, are filling a role that private VC structurally cannot.[s]
Geopolitical Calculus
The sovereign wealth fund AI pivot carries geopolitical weight that transcends financial returns. The Berkeley Political Review observed that SWFs “challenge the traditional separation between markets and state power” because capital allocation becomes “a mechanism through which governments can exert influence beyond their borders.”[s]
For Gulf states, the logic is multi-layered. By providing essential compute infrastructure to American AI companies, they create mutual dependency that functions as diplomatic insurance. The Stargate Project exemplifies this: U.S. sovereignty over AI model development theoretically increases because training happens on American soil, but the capital enabling that sovereignty flows partly from Abu Dhabi’s MGX.[s] Washington’s loosening of chip export controls to the Gulf, which HUMAIN’s CEO expects to continue[s], reflects the leverage these investments create.
The model also carries risks. The Committee on Foreign Investment in the United States (CFIUS) has expanded oversight of foreign acquisitions involving critical technologies including AI, and similar frameworks have emerged across Europe, Japan, and Australia.[s] Sovereign AI investments sit at the intersection of economic partnership and security concern.
The Norwegian Counterpoint
Norway’s Government Pension Fund Global offers an instructive contrast. At $2 trillion, the world’s largest sovereign wealth fund earned $247 billion in 2025, with its annual report citing Nvidia, Alphabet, and Broadcom as the largest contributors to that return.[s] Rather than building AI infrastructure directly, Norway benefits from AI adoption as a passive equity holder in the companies building it.
Internally, the fund has become an AI adopter: since November 2024, it has used large language models to screen every new equity portfolio addition within 24 hours for ESG risks including forced labor, corruption, and fraud.[s] The system flags information “not captured in international media coverage or data vendor alerts,” and in multiple instances the fund sold positions before markets priced in the risk.[s] Norway’s approach demonstrates that sovereign wealth fund AI strategy need not mean building infrastructure; it can also mean deploying AI as an analytical edge within existing investment operations.
Sovereign AI Beyond the Gulf
The phenomenon extends globally. France announced roughly €109 billion in AI-related investment commitments from private and international capital, while France 2030 contributes public AI and R&D funding within that broader ecosystem; the 1.2 million GPU target is best attributed to Sesterce’s French AI infrastructure plan rather than a standalone official government goal.[s] Global sovereign AI spending is projected to surpass $100 billion by 2026.[s] The U.S. itself moved toward a sovereign wealth fund: in February 2025 President Trump signed an executive order directing officials to develop a plan for establishing one, while separately the federal government pursued direct equity stakes in strategic semiconductor ventures including Intel and the xLight laser startup.[s]
Risk Assessment
The concentration of sovereign capital in AI infrastructure creates exposure on multiple fronts. Norway’s fund CEO Nicolai Tangen has publicly identified an AI bubble as a major risk scenario. Regulatory fragmentation (differing chip export rules, data sovereignty requirements, CFIUS-style reviews) could balkanize the global AI infrastructure market that these investments assume will remain interconnected.
For Gulf states, the infrastructure itself may represent an asymmetric hedge: even if AI startup valuations correct, gigawatt-scale data centers remain physical assets generating revenue from cloud compute demand. The PIF’s $700 billion portfolio can absorb volatility that would devastate a smaller allocator. Whether the sovereign wealth fund AI thesis ultimately validates depends less on any single investment and more on whether compute demand sustains its current growth trajectory through the end of the decade, a question that will determine whether these oil-era fortunes successfully crossed the bridge to the digital economy.
