Everything About AI
NextFin News - Humain, the Saudi artificial-intelligence company owned by the Public Investment Fund, is planning to raise an initial $2.5 billion from global and domestic investors for a fund dedicated to building data centers in the Kingdom, according to people familiar with the matter. The fund would finance 250 megawatts of data-center capacity that Humain is developing with Al Moammar Information Systems Co., with the pipeline potentially expanding to as much as 1 gigawatt.
The move marks a pivot in how Saudi Arabia pays for its AI ambitions. For the first 16 months of Humain's existence, the financing model was straightforward: the state writes the checks. Now the Kingdom is testing whether outside capital will underwrite the concrete, steel and gas turbines behind the announcements. The question is not whether Saudi Arabia wants AI infrastructure - it does, with a national strategy targeting 1.5 gigawatts of data-center capacity by 2030. The question is whether investors will fund it at the pace the state envisions, and who absorbs the risk when power connections slip.
The Deal, the Partner and the Mandate
Humain was launched in May 2025 under the Public Investment Fund, chaired by Crown Prince Mohammed bin Salman in his capacity as PIF chairman, with a mandate to build the entire AI stack: data centers, cloud infrastructure, models and applications. In October 2025, PIF and Aramco signed a non-binding term sheet under which Aramco would acquire a significant minority stake in Humain, with PIF retaining majority ownership. Both parties would contribute AI assets, capabilities and talent into the company, and the stated intention was to enable the rapid scaling up of Humain's operations to capture value in the AI sector.
The new fund sits on top of a crowded stack of previously announced commitments. Humain and Australia's AirTrunk, backed by Blackstone, agreed in late 2025 to develop a data-center campus in the Kingdom in an initial project of approximately $3 billion. Earlier this year, the Saudi Infra Fund and Humain agreed on non-binding financing terms of up to $1.2 billion for as much as 250 megawatts of AI data-center capacity, according to a company statement. And in August, Al Moammar Information Systems - known as MIS - said it signed a five-year contract to deliver data-center hosting services to Humain, a deal it said was worth more than 30% of its total revenue in fiscal 2025 and would run from the third quarter of 2026 through the third quarter of 2031.
MIS is not a marginal partner. The Riyadh-listed technology company said before the Humain agreement that the value of its ongoing projects reached SAR 10 billion, and it has set a target of capturing 20% of the Saudi data-center development market. According to an initiation report by ANB Capital, MIS had deployed 32 megawatts of a planned 120 megawatts of data-center capacity as of May 2025. The company reported revenue of SAR 1.27 billion in fiscal 2025. The pairing makes strategic sense: Humain brings the sovereign mandate and the anchor-tenant pipeline; MIS brings a listed balance sheet, local delivery capability and a track record of building Tier III facilities in Dammam and Riyadh.
The pairing also reflects a broader pattern. Humain's mission, as described by the sovereign fund, includes building next-generation data centers with computing power incorporating hundreds of thousands of NVIDIA's most advanced GPUs, cloud platforms, and advanced AI models such as ALLAM, a multimodal Arabic large-language model co-developed with the Saudi Data and Artificial Intelligence Authority. Partnerships span NVIDIA, Microsoft, AMD, Qualcomm, Amazon Web Services, Google Cloud and groq. In a 2025 interview, Humain's chief executive outlined a plan costing around $77 billion at then-current market rates to deliver the full build-out.
Why a Fund, and Why Now?
The most important detail in this story is not the $2.5 billion figure. It is the structure. A fund changes the discipline. Third-party investors - global pension funds, sovereign wealth funds, regional family offices - do not write checks on the basis of national strategy. They underwrite contracted cash flows, power purchase agreements, interconnection dates and tenant credit. By seeking external capital, Humain is effectively inviting the market to price Saudi data-center risk.
If the fund closes at or near the reported $2.5 billion target, it would signal that international investors are willing to accept Saudi power-sector execution risk at a price. If it closes small, or slowly, it would tell a different story about how much of the AI build-out the state will ultimately have to carry itself. That is the real referendum embedded in the fundraising.
There is also a balance-sheet logic on the Aramco side. Amin H. Nasser, Aramco's president and chief executive, said at the time of the stake agreement:
Aramco is well positioned to capture opportunities from rising energy demand linked to AI growth.
A fund structure lets Aramco and PIF recycle capital: they can seed projects, bring in co-investors once the power and tenant risk is de-risked, and redeploy the proceeds into the next site. That is the classic infrastructure-general-partner playbook, and it is a more mature operating model than the announcement-driven approach of 2025.
Yazeed A. Al-Humied, deputy governor and head of MENA investments at PIF, framed the combination of the two state giants in similar terms:
By combining PIF and Aramco's AI assets under HUMAIN, we are fueling AI talent, innovation and intellectual property, while aligning and accelerating future investment opportunities.
The Real Constraint Is Power, Not Capital
Here is the uncomfortable number behind the $2.5 billion. Saudi Arabia operated 467 megawatts of data-center capacity in the first quarter of 2026, the communications ministry reported - an increase of more than 6% since the start of the year, and up from 68 megawatts in 2021. The sector has expanded nearly sixfold since the launch of Vision 2030, with more than 60 operational facilities developed by more than 20 companies.
Against that installed base, the targets are staggeringly larger. The National Data Center Strategy, issued by the communications ministry with the Saudi Data and Artificial Intelligence Authority in June 2025, sets a goal of about 1.5 gigawatts by 2030 - roughly 3.2 times what is energized today. Humain's own corporate pipeline runs further still: the company's chief executive has described a plan for 6.6 gigawatts by 2034, which is roughly 14 times the current operational base.
Chips get the headlines. Grids get the delays. A data center without a power interconnection is an expensive shell, and the bottlenecks are physical: gas-turbine procurement lead times, transmission build-out, cooling-water access and energization queues. These are not problems that a $2.5 billion fund solves on their own. They are problems that require coordination between the energy ministry, the grid operator, water authorities and the developers - and they typically resolve on multi-year timelines, not quarterly fundraising cycles.
This is where the MIS partnership matters more than it first appears. MIS's model - build, own and operate smaller Tier III facilities distributed across multiple cities - is less glamorous than a single gigawatt-scale AI campus, but it is also less exposed to the single-point grid-connection risk that can stall a mega-project for years. A portfolio of 48-megawatt and 250-megawatt sites, each with its own interconnection path, diversifies execution risk in a way that a handful of colossal campuses do not. The fund's 250-megawatt first tranche is a meaningful step in that direction, but it is also, on its own, only about half of one percent of the 6.6-gigawatt pipeline.
The market is growing fast enough to absorb new supply - for now. The Saudi data-center construction market was valued at $1.61 billion in 2025 and is estimated to reach $8.11 billion by 2031, a compound annual growth rate of 30.91%, according to Mordor Intelligence. But growth rates this high are exactly what attract overbuilding, and the history of infrastructure cycles is that the last projects commissioned before demand pauses are the ones that sit underutilized longest. The Kingdom's advantage - abundant land, competitive energy pricing and available power relative to more mature markets - is real, but it is a comparative advantage, not a guarantee against mis-timing.
Full-Stack Ambition Meets Real-Estate Economics
Humain's mandate is deliberately broad, and that breadth creates a capital-allocation tension that every diversified technology conglomerate eventually has to answer. A data center is a long-duration, utility-style asset: high upfront capex, contracted revenues, returns driven by power costs and utilization. An AI model business is a winner-take-most software race: burning cash on compute and talent today for a chance at platform economics tomorrow. When power is scarce, does the marginal megawatt go to the highest-paying colocation tenant, or to the internal model-training cluster whose payoff is years away?
The fund structure partially resolves this by ring-fencing the infrastructure. Investors in the data-center vehicle are buying real estate and power, not optionality on ALLAM's adoption curve. That is cleaner for pricing, but it also means Humain must find separate capital for the model and applications layers - or subsidize them from sovereign balance sheets. The state has shown it is willing to do the latter; the question is whether the infrastructure fund's returns can stand on their own.
There is a second-order implication here that the market has not fully priced. If Saudi Arabia succeeds in building out low-cost, gas-fired power for AI data centers at scale, it does not just create a regional hosting hub - it changes the global economics of model training. Cheap, dedicated power is the single largest operating-cost lever in AI compute. A jurisdiction that can deliver it reliably, at a discount to the grids of Northern Virginia or Northern California, can attract training workloads even if latency is suboptimal, because training is largely delay-tolerant. That is the structural prize: Saudi Arabia is not competing to be the fastest data center; it is competing to be the cheapest kilowatt-hour.
But that prize depends on execution, and execution is where the plan meets the grid.
The Counter-Thesis: A Fund Does Not Change the Arithmetic
The strongest argument against reading too much into this fund is simple: the money was never the binding constraint. PIF, one of the world's largest sovereign wealth funds, and Aramco, the world's largest oil exporter, can fund far more than $2.5 billion if they choose to. A 250-megawatt tranche is marginal against a 6.6-gigawatt pipeline. If the state wants the capacity built, it will build it - with or without a fund. From this angle, the fund is a sideshow: a public-relations exercise in "crowding in" private capital that changes neither the physics of the grid nor the timeline of the build-out.
There is force in that view. Saudi Arabia's AI strategy has never lacked for announced capital; it has lacked for energized megawatts. And a fund that prices Saudi risk correctly might simply decline to invest, leaving the state to step in anyway. The counter-thesis also warns of a familiar emerging-market pattern: infrastructure announced in tranches, delivered in dribs, with the gap between the two absorbed by the public balance sheet.
But the counter-thesis misses what the fund is actually doing. It is not primarily about raising money; it is about creating a pricing signal and an operating discipline that balance-sheet funding cannot provide. A sovereign check does not force a developer to lock in a power purchase agreement, secure an interconnection date, or sign a tenant before breaking ground. Fund investors do. If Humain can close this fund and deploy it against contracted, energized capacity, it would demonstrate a replicable model for the remaining gigawatts. If it cannot, the state will indeed carry the load - but at the cost of slower deployment and weaker capital discipline.
The falsifying signal is concrete. Watch two things: the fund's closing size and the energization dates of the first 250 megawatts. If the fund closes materially below $2.5 billion, or if the 250 megawatts are not energized within 36 months of the fund's close, the thesis that third-party capital can accelerate Saudi AI infrastructure is wrong - and the build-out will revert to the slower, state-funded path that has defined the sector to date.
What to Watch Next
In the short term, the signal is fundraising: how much capital Humain commits, from whom, and on what terms. A close near the $2.5 billion target with global institutional investors would be a stronger endorsement than a close funded largely by regional related parties.
Over the medium term, the metric that matters is megawatts energized, not megawatts announced. The 250-megawatt tranche with MIS is the first real test. Track the interconnection agreements, the power procurement structure, and whether the hosting contract signed in August begins contributing to MIS revenue in the third quarter of 2026 as the company guided. MIS reported first-half 2026 revenue of SAR 910.1 million, up 28.2% year on year, so the Humain contract - worth more than 30% of its full-year 2025 revenue - should be visibly accretive if it lands on schedule.
In the long term, the question is whether Saudi Arabia can convert its capital advantage into a durable AI hub, or whether cheap power and deep pockets prove necessary but not sufficient. The state has the money. It has the land. It has the political will. What it still needs to prove is that it can connect the plugs to the sockets faster than the rest of the world can build the chips.
The bottom line: Humain's $2.5 billion fund is less a financing event than a referendum on whether global capital will underwrite Saudi Arabia's AI infrastructure at the speed the Kingdom wants it built. The money is the easy part. The megawatts are not.
Explore more exclusive insights at nextfin.ai.
