Human expands its partnership with Nvidia to 84 megawatts, allocating 50 megawatts to an American client
Saudi company Human announced the expansion of its partnership with Nvidia to develop an AI computing infrastructure of 84 megawatts, comprising about 27,000 advanced GPUs, revealing a notable shift in the regional data-center model, with 50 megawatts of future capacity earmarked for a unnamed American client. The announcement was made at the GTC conference in Berlin, confirming the kingdom’s accelerated plans to become an international supercomputing hub, leveraging its ownership by the Public Investment Fund and the chairmanship of Crown Prince Mohammed bin Salman of its board.
The new computing capacity is divided into two operational phases reflecting the progression of technology generations;The first phase launches in the fourth quarter of the current year with 14 megawatts operating on Blackwell Hgx300 systemsThis equates to roughly 6,000 GPUs dedicated to Human Cloud customers from government agencies, commercial firms and AI developers. The second phase begins in mid-2027 with the addition of 70 megawatts of next-generation Vera Robin processors, representing about 21,000 GPUs, and includes operating 50 megawatts for the American client starting in June, followed by an additional 20 megawatts for the company’s local cloud between July and August, while Robin projects remain subject to final agreements and implementation schedules.
The expansion continues a trajectory that began in May 2025 during U.S. President Donald Trump’s visit to Riyadh, when Nvidia unveiled a plan to supply the company with hundreds of thousands of advanced chips over five years, including an initial batch of 18,000 Blackwell processors as part of a phased target to reach 500 megawatts. Human, led by CEO Tarek Amin, who affirmed the aim of bringing the latest accelerated computing generations to the kingdom at a significant scale, seeks to reach 3 gigawatts of data-center capacity by 2030 and exceed 6 gigawatts by 2034.
The move places Riyadh in direct competition with Abu Dhabi, where companies G42, OpenAI and Oracle are developing the Emirates StarGate complex with an initial capacity of one gigawatt as the first stage of a computing campus targeting 5 gigawatts in partnership with Nvidia. The Gulf nations rely critically on U.S. export approvals to secure advanced chips, making Washington’s evolving policy and technical restrictions the key factor in determining the pace of regional expansion.
This achievement translates into practical impacts on the Gulf’s tech business environment through two pathways;The first is providing an advanced local cloud infrastructure that reduces regional software teams and institutions’ reliance on outsourced computingThis shortens response times and addresses the need to localize sensitive data for financial and governmental sectors. The second is the entry of Gulf computing into a reverse-export market of compute capacity to Western firms, which compels regional technology leaders and development teams to accelerate alignment of their systems with Blackwell and Vera Robin architectures, while preparing for possible competition for cloud capacity shares amid growing international demand.
The Berlin announcement coincided with Saudi Arabia’s postponement of the Future Investment Initiative forum to next year, following the October 10 attack on King Khalid International Airport in Riyadh that killed 13 people and injured more than 300. It shows that the continued signing of major deals at global venues underscores Riyadh’s focus on advancing its sovereign AI capabilities according to its set timelines.