How Gulf Funds Are Buying The Backbone Of AI

Sovereign wealth funds from the Gulf have moved beyond buying AI stocks. They're buying the buildings, the power plants, and the debt that the entire AI economy runs on — and the scale of that bet should make anyone paying attention at least a little nervous.

Gulf state funds deployed roughly $66 billion into artificial intelligence and digital infrastructure during 2025, according to Global SWF's annual report, in a year when their combined assets reached a record $15 trillion. The seven largest Gulf funds accounted for 43% of all sovereign capital invested globally — some $126 billion across every sector. Abu Dhabi's Mubadala alone committed $12.9 billion to AI and digital deals. Saudi Arabia's Public Investment Fund put $36.2 billion across AI-related transactions, much of it tied to a single large acquisition.

The numbers are staggering. But the direction of travel matters more than the totals.

## From Passive Investor to Infrastructure Owner

The shift underway is fundamental. These funds aren't buying shares in Nvidia and waiting for them to appreciate. They're going upstream — into the physical infrastructure that AI runs on, the compute that every model needs, and the power that keeps the data centers humming.

The clearest example came in October 2025, when a consortium of BlackRock's Global Infrastructure Partners, Abu Dhabi's MGX, and the AI Infrastructure Partnership agreed to acquire Aligned Data Centers in a deal implying an enterprise value of roughly $40 billion, covering more than 50 campuses across the Americas. That's not a venture investment. That's owning the ground the AI economy is built on.

Saudi Arabia's National Infrastructure Fund and Humain, an AI company backed by the Public Investment Fund, followed up in January 2026 with a financing framework worth up to $1.2 billion to build as much as 250 megawatts of data-centre capacity in the kingdom. The structure of these commitments increasingly resembles infrastructure finance — long horizons, contracted revenue from hard assets, steady yields. This is a world away from the venture-stage AI bets of 2022 and 2023.

The calculation is straightforward: picking which AI model wins is a hard bet. Owning the data centers, fiber, and power that every model needs is a steadier one. Compute demand grows regardless of which company comes out on top. And for Gulf states facing a future where oil revenue becomes less central to their economies, AI infrastructure offers both a financial return and a measure of technological standing on the global stage.

## The Concentration Problem

Here's what should give everyone pause. A fund that holds AI equities, owns the data centers those companies lease, and finances the power feeding them has effectively made the same bet three times over — on the continued growth of compute demand. Should that demand disappoint, the losses would arrive across all three holdings together, not one by one.

The fashionable term for this is "patient capital," and there's truth in it. Gulf sovereign funds have investment horizons measured in decades, not quarters. They can ride out a downturn that would force a leveraged private equity buyer to sell at the worst possible time. They're buying real assets with contracted income, a long way from speculative software plays.

But the less comfortable reading is concentration. The Financial Stability Board, the G20's risk monitor, warned in May that heavy exposure to AI infrastructure could leave investors facing sizeable losses if a glut of data centers outpaced demand for compute. Sovereign funds aren't private-credit vehicles, but the underlying hazard is the same: a single thesis, expressed through many instruments, that has never been tested by a downturn in the technology it depends on.

## Why This Matters Beyond the Gulf

The Gulf's AI infrastructure play has implications that reach well beyond Abu Dhabi and Riyadh.

**For the AI industry:** Sovereign capital is patient, but it's also strategic. When a single region owns a growing share of the physical infrastructure that AI runs on, it acquires leverage over the companies that depend on that infrastructure. Lease negotiations, data sovereignty requirements, and regulatory compliance all become more complicated when your landlord is a sovereign wealth fund with a 30-year time horizon and geopolitical objectives.

**For data sovereignty:** AI data centers process enormous volumes of sensitive data — health records, financial transactions, military intelligence. When those data centers are owned or financed by foreign governments, it raises questions about data access, jurisdiction, and national security that haven't been adequately addressed by existing frameworks.

**For competition:** If compute becomes concentrated in facilities owned by a small number of deep-pocketed sovereign investors, it could create barriers to entry for new AI companies that need affordable access to training infrastructure. The opposite of the democratization that cloud computing was supposed to deliver.

**For investors:** The concentration risk extends beyond the Gulf funds themselves. Anyone invested in AI infrastructure — whether through data center REITs, semiconductor stocks, or power utilities serving data centers — is making a variant of the same bet. If compute demand disappoints, the entire ecosystem feels it simultaneously.

## The Uncomfortable Question

No one is saying the Gulf funds are wrong about AI's growth trajectory. The demand for compute is real, and it's growing faster than most analysts predicted even two years ago. The infrastructure buildout is necessary, and patient capital is exactly what's needed to finance it.

But there's a difference between making a bet and staking your entire strategy on it. The Gulf has committed a remarkable amount of capital to a single thesis — that AI compute demand will continue to grow, that data centers will remain fully utilized, and that the returns on infrastructure will justify the enormous upfront investment. It's a plausible thesis. It may well be correct. But it has never been stress-tested by a real downturn in AI demand, and until it has, the concentration risk is real.

The AI industry's brief history is littered with cycles of boom and bust. The question isn't whether compute demand will grow over the long term. It almost certainly will. The question is whether it will grow smoothly enough to support the infrastructure that's being built on the assumption that it will — and what happens to the global financial system if it doesn't.

## What This Means For You

**If you're invested in AI stocks or infrastructure:** Understand that you may be making a more concentrated bet than you realize. If you own semiconductor stocks, data center REITs, and power utilities serving AI facilities, you're effectively betting on the same outcome three ways. Diversification across sub-sectors doesn't help if the underlying thesis is the same.

**If you work in tech or AI:** The Gulf's infrastructure play means more data centers will get built, which should lower compute costs over time — good for startups and researchers. But it also means a growing share of the physical backbone of AI is owned by entities with strategic interests that extend beyond financial returns. Pay attention to where your compute comes from.

**If you're a policymaker or work in government:** The data sovereignty implications are urgent. Current regulatory frameworks weren't designed for a world where foreign sovereign wealth funds own the data centers processing your citizens' sensitive information. The time to develop clear rules about data residency, access, and jurisdiction is now, not after the next geopolitical crisis makes it impossible to ignore.