Saudi company HUMAIN plans an expanded NVIDIA partnership that would add up to 84 megawatts of AI computing capacity, about 27,000 GPUs, as it broadens access to advanced computing.
HUMAIN said the expansion is meant to widen access for enterprises, developers and AI innovators. Its public announcement gave no contract value, customer names, financing terms or precise commissioning dates.
The National reported that the first phase would comprise 14 megawatts of Blackwell HGX B300 systems, or about 6,000 GPUs, followed by a 70-megawatt Vera Rubin phase with roughly 21,000 GPUs. The outlet said 50 megawatts of the Vera Rubin allocation would go to an unnamed U.S. customer, with the remaining 20 megawatts for HUMAIN Cloud.
The National put the U.S. project's target at June 2027 and HUMAIN Cloud's at July or August 2027. It reported that both Vera Rubin projects were subject to final agreements and deployment plans.
Tareq Amin, HUMAIN's chief executive, told The National: “We are moving quickly to bring the next generations of Nvidia-accelerated computing into Saudi Arabia at meaningful scale.” NVIDIA’s product sheet lists HGX B300 as part of its Blackwell Ultra platform.
NVIDIA said in May 2025 that HUMAIN, a company of Saudi Arabia’s Public Investment Fund, planned up to 500 megawatts of AI factories over five years, including an initial 18,000-GPU GB300 Grace Blackwell supercomputer. The announcement also described infrastructure for sovereign AI models. In November 2025, HUMAIN announced a three-year plan involving up to 600,000 NVIDIA GPUs or AI infrastructure technologies across Saudi Arabia and the United States.
HUMAIN’s public post did not say whether the 84-megawatt expansion counts toward its earlier targets or adds capacity beyond them. The company plans to begin Blackwell Ultra deployments in the fourth quarter of 2026, with Vera Rubin deployments planned for 2027.
