HUMAIN and KAUST plan to offer selected researchers and enterprises access to Shaheen III’s CPU computing capacity through HUMAIN Cloud. (Shutterstock)Saudi Arabian artificial intelligence company HUMAIN and King Abdullah University of Science and Technology are working toward making part of KAUST’s Shaheen III supercomputer accessible through HUMAIN Cloud. The proposed arrangement would place the system’s CPU-based computing capacity alongside the cloud platform’s planned AI tools, models and developer services.
The collaboration is still at a preliminary stage. HUMAIN and KAUST have not announced a definitive commercial supply agreement, pricing structure, launch date or volume of capacity that would be offered to outside users.
If completed, the project would give selected companies and researchers another route to computing resources normally associated with large scientific institutions. Intended workloads include data analytics, scientific discovery, engineering simulations and digital twins. Biotechnology, materials science, oil and gas, and manufacturing are among the fields identified by the two organizations as possible areas of collaboration.
Tareq Amin, CEO of HUMAIN, said: "The quality, scale, and accessibility of the infrastructure behind it will shape the next era of AI. HUMAIN Cloud is designed to bring together advanced compute, AI platforms, frontier models, and developer services into a unified environment that enables innovation at scale. Working with KAUST to bring Shaheen III CPU capability into the HUMAIN Cloud ecosystem would expand the capabilities available to researchers, developers, and enterprises, creating new opportunities to accelerate AI development and solve increasingly complex computational challenges at unprecedented scale."
Shaheen III is not a single, uniform computing cluster. The HPE Cray EX system has separate CPU and GPU-accelerated partitions, allowing KAUST to assign different types of hardware to different workloads.
KAUST says the CPU system contains 4,608 compute nodes using AMD Genoa processors. Each node provides 192 CPU cores and 384 gigabytes of memory, while the partition delivers 35.6 petaflops of sustained performance under the High Performance Linpack benchmark. It ranked 64th on the June 2026 TOP500 list. These specifications make the system suitable for processor-intensive modeling, large-scale data analysis and other scientific workloads that do not necessarily depend on AI-focused graphics processors.
A second partition contains 2,800 Nvidia GH200 Grace Hopper Superchips and has recorded 122.8 petaflops of sustained benchmark performance. KAUST’s technical description of Shaheen III identifies that GPU system as the Middle East’s fastest supercomputer at the time of installation. The university originally selected Hewlett Packard Enterprise to build Shaheen III as a successor to its earlier computing platform, with HPE saying the system was designed to support research in areas including food, water, energy and the environment.
KAUST will retain sole ownership and operational control of Shaheen III under the proposed structure. Existing computing allocations for university researchers are expected to remain unchanged. The university has also said the system operates under applicable export-control rules and license conditions, meaning access would not simply be open to every prospective cloud customer. Eligibility, permitted workloads and regulatory compliance would all influence who could use the service.
The arrangement could help HUMAIN add established CPU computing capacity while it develops a much larger cloud and data-center portfolio. The Public Investment Fund-owned company was established in 2025 to operate across AI infrastructure, cloud services, models and applications. HUMAIN has since pursued hardware and infrastructure relationships with companies including Nvidia, AMD and Cisco. In August, AMD and Cisco said their first production infrastructure for HUMAIN had gone live in Saudi Arabia, with further capacity planned from 2027.
Building out advanced computing infrastructure at this scale will require sustained investment, reliable power, specialized hardware and long-term construction planning. HUMAIN’s expansion also depends on securing enough advanced processors and bringing new capacity online as scheduled. The KAUST collaboration could provide useful computing resources during that wider buildout.
Connecting Shaheen III to a cloud platform would represent more than a branding exercise. Supercomputers are typically managed as tightly controlled institutional resources, with access awarded through research programs or dedicated partnerships. Delivering CPU capacity through HUMAIN Cloud could place a commercial interface between the underlying machine and approved users, potentially making specialized computing easier to request and integrate into company workflows.
Professor Sir Edward Byrne AC, President of KAUST, said: "As demand for HPC and AI infrastructure continues to grow, the collaboration reflects a shared commitment to making advanced computing more accessible. By leveraging leading HPC capabilities through HUMAIN Cloud, HUMAIN and KAUST are creating a platform to accelerate scientific discovery, AI innovation, and the development of next-generation intelligent applications."
HUMAIN and KAUST also plan to examine technical blueprints and reference architectures for industry-specific applications. Such materials could help organizations configure software for demanding simulations or analytical tasks without designing every part of the computing environment themselves.
