HUMAIN’s CEO Tareq Amin frames Saudi Arabia’s recent moves to rewire its economy around artificial intelligence as a bold bet: the kingdom aims to join the United States and China among global AI superpowers.
Amin, a Jordanian‑American tech executive who previously led the technology arm of Saudi Aramco, helped launch the state‑backed AI company ahead of US President Donald Trump’s visit to the Kingdom last year. The concept leverages Saudi Arabia’s low‑cost energy, large land availability and sovereign capital to build one of the world’s largest compute hubs.
Past deals and expansion plans
Amin recorded an early win with a timely $3 billion investment into Elon Musk’s xAI before that business was folded into SpaceX ahead of a historic IPO. HUMAIN now plans to create a large venture capital fund and expand with offices in the United States, France and the United Kingdom to “accelerate the pace of investment.”
The company also wants to reduce reliance on the kingdom’s sovereign wealth fund, the Public Investment Fund (PIF), which is its largest shareholder, and to open ownership to more global investors. Amin describes HUMAIN’s intended role as a kind of “Switzerland” for open models, a neutral marketplace that could avoid direct confrontation between US and Chinese AI ambitions.
Ownership targets and financing mix
Amin says his target ownership split is 30% shareholder financing and 70% third‑party capital. While that ratio has not been reached, he says they are close and that opening HUMAIN to the global investment community is a priority.
HUMAIN’s business model combines several elements: build‑to‑suit data centers for hyperscalers (who purchase the GPUs) and AI inferencing services — endpoint APIs to sell frontier models on a marketplace. That approach is capital‑intensive.
Amin says major external commitments are already in place: Blackstone has committed $3 billion, and other large firms are expected to participate to accelerate HUMAIN’s debt and equity raise.
Compute targets and costs
Citing industry averages, Amin said a typical build cost is about $11–12 million per megawatt for data centers. HUMAIN has declared a build target of 3 gigawatts; using industry figures, that scale implies multi‑billion‑dollar — even tens‑of‑billions‑of‑dollars — capital needs (the reporting notes that this would be more than $30 billion by some calculations). Amin did not give a precise total.
He also stated that much of HUMAIN’s capacity is sold before it is deployed, which he takes as evidence that the market is moving from experimentation to real enterprise AI implementation.
Sovereign AI, partnerships and model strategy
Recent US‑level restrictions on access to some models have sharpened interest in the concept of sovereign AI. Amin argues there are two key constraints to building frontier models: deep scientific talent and very large compute resources.
HUMAIN has a strategic partnership with a US firm called Reflection AI. Amin says the relationship both helps train HUMAIN’s team in tuning, optimizing and improving Allam (the company’s Arabic‑first large language model) and provides the sovereignty they require. Amin also highlighted a major announcement with Mistral and said HUMAIN wants to be a marketplace — a neutral hub — for open models. He believes open‑weight models are increasingly reaching frontier performance.
As an illustration of the compute gap, Amin says Allam used about 4,000 GPUs in its development, compared with OpenAI’s use of “hundreds of thousands” of GPUs. That difference underscores the compute challenge for any group attempting to build state‑of‑the‑art frontier models.
Bloomberg reported that HUMAIN’s AI platform is built on MiniMax’s M3, a Chinese open‑source model, as part of its push to be globally competitive in the AI race.
Balancing build speed and revenue offtake
Amin describes HUMAIN’s approach as a balancing act between accelerating construction and ensuring predictable revenue offtake from customers, which is critical for securing third‑party debt and equity. He said he is not worried about land availability or grid capacity, but stresses the need for a predictable relationship between offtake agreements and build schedules.
He also warned that global power capacity may be insufficient for the expected surge in compute demand, and that HUMAIN is preparing contingencies. If asked by “His Royal Highness” to accelerate, Amin said the company has the levers to do so, but for now prefers a measured strategy linking offtake and build.
Why it matters
HUMAIN’s plans combine large‑scale infrastructure, international partnerships and a marketplace approach to position Saudi Arabia as a serious player in global AI infrastructure. Given the costs and the early commitments — such as the Blackstone $3 billion commitment and the stated 3 GW build target — the project signals a multi‑billion‑dollar push to secure sovereign AI capacity and attract global investors.
