AI News
Aramco Closes $2.6 Billion Deal for 49% Stake in Humain, Saudi Arabia's AI Company
Saudi Aramco has closed its deal for a 49% stake in Humain, the artificial intelligence company backed by Saudi Arabia's Public Investment Fund (PIF). The closing was announced on November 1, with Aramco holding the stake through a wholly owned subsidiary and the PIF retaining the remaining 51%.
Under the deal, Aramco will contribute about SAR 9.9 billion, roughly $2.6 billion, in cash during 2026, alongside AI, computing and digital assets. The transaction formalises a non-binding term sheet the two sides signed on October 28, 2025, turning a year-old outline into a completed investment.
Humain was launched in May 2025 as a PIF company with four lines of business: data centres, high-performance computing and cloud services, AI models, and AI solutions for businesses and governments. Its best-known model work is ALLaM, a multimodal large language model built for Arabic. The company sits at the centre of Saudi Arabia's plan to build a domestic AI industry rather than rent one from abroad.
The headline target attached to the deal is 3 GW of AI data-centre capacity by 2030. That is an ambitious scale by any measure: a single gigawatt of data-centre capacity is a very large facility, and three of them would put Humain's programme among the biggest AI infrastructure build-outs announced anywhere. But the announcements so far describe a target, not a construction schedule.
What is not disclosed matters as much as what is. No valuation for Humain has been published, and no value has been put on the AI, computing and digital assets Aramco is contributing. There is no payment schedule beyond the cash arriving during 2026, and no sites, power sources or capital spending figures for the 3 GW have been named. The $2.6 billion is a shareholder contribution to Humain, not a construction budget for data centres, and readers should not treat it as one.
A separate, earlier step hints at where the first capacity may come from. On October 21, Humain and DataVolt announced a preliminary framework with Nebius for up to 50 MW of capacity at the Oxagon campus in NEOM, part of an initial 100 MW already under construction with a first phase of 360 MW, phased from the fourth quarter of 2027. That framework is preliminary and non-binding, so it is a signal of intent, not a signed build.
For Aramco, the logic is straightforward. The world's largest oil producer has spent years applying AI to its own operations, and its chief executive, Amin Nasser, has cited $11.3 billion in Technology Realized Value since 2023, a significant portion of it from AI. An energy company with cheap power, capital and industrial sites is a natural partner for an AI build-out whose binding constraint is often electricity, not chips.
What to watch next is execution. The cash contributions land during 2026, so the first test is whether Humain announces concrete data-centre sites, power arrangements and partners to match the 3 GW target. The second is whether the 3 GW counts only Humain-owned capacity or includes partner capacity such as the Nebius framework. The third is progress on the model and solutions side, where ALLaM and enterprise products have to prove themselves against global competitors.
The deal also fits a wider Saudi pattern. Under the Vision 2030 programme, the kingdom has made AI and data infrastructure a national priority, and the PIF has become the main vehicle for building it. Humain, launched only in May 2025, is the PIF's dedicated AI company — and an Aramco investment gives it something a fund alone cannot: an industrial partner with power, sites and operating experience, plus a balance sheet that can fund a multi-year build.
For Aramco shareholders, the size of the commitment is modest against the company's overall capital spending, which runs into the tens of billions of dollars a year. The risk is not the money; it is whether Humain can convert a stake sale and a capacity target into working data centres, customers and models. The $2.6 billion buys a seat at the table. The 3 GW, if it is built, is what would make the seat worth having.
The deal itself is no longer a proposal: it closed on November 1. The open questions are now practical ones about building, power and products, and they will decide whether this becomes a working AI industry or an expensive announcement.