Global AI achieved NVIDIA's reference-architecture performance targets on frontier-model training workloads at 512-GPU scale on NVIDIA GB300 NVL72 systems.

‍

ENDICOTT, NY – 07/01/26:  Global AI today announced that it has successfully achieved NVIDIA Exemplar Cloud validation on NVIDIA GB300 NVL72 systems for training workloads, demonstrating that dedicated sovereign AI infrastructure can deliver frontier scale AI performance while maintaining complete customer control over data, infrastructure, and operations.

The NVIDIA Exemplar Cloud validation provides independent confirmation that Global AI’s sovereign infrastructure environment can operate at the performance levels required for advanced AI training while preserving the physical, operational, and governance controls increasingly required by regulated industries, and enterprises operating mission-critical AI workloads.

‍

‍

“We founded Global AI around a simple conviction: organizations should be able to run advanced AI within environments they control. The question was never whether sovereignty mattered. The question was whether it could be delivered at the scale, performance, and operational complexity that AI demands. This validation is an important proof point that it can.” Said Sami Issa, CEO and Co-Founder, Global AI.

‍

During the NVIDIA Exemplar Cloud validation process, Global AI achieved NVIDIA reference architecture performance targets across frontier-model training workloads running on NVIDIA GB300 NVL72 infrastructure at 512-GPU scale.

The benchmarking process validated not only hardware performance, but also the operational maturity of Global AI’s software platform, orchestration capabilities, networking architecture, cooling systems, and infrastructure management practices required to consistently support large-scale AI training environments.

The NVIDIA benchmarking recipes were executed through Global AI’s proprietary software platform built on a SLURM-based orchestration framework. The validation demonstrated the company’s ability to manage, optimize, and operate complex distributed training workloads across advanced NVIDIA infrastructure while maintaining the isolation, security, and operational control required for sovereign AI deployments.

‍

Unlike traditional cloud environments, Global AI delivers dedicated, single-tenant infrastructure where customers maintain full control of their data, workloads, and operational environments. The validation reinforces the company’s vision that sovereign AI infrastructure should combine the control traditionally associated with on-premises environments with the performance, reliability, and operational sophistication of the world’s leading hyperscalers.

‍

“Working alongside NVIDIA through the NVIDIA Exemplar Clouds initiative allowed us to push the limits of performance on GB300 NVL72. The level of collaboration and rigor involved in the process validated the strength of our platform, our engineering organization, and the efficiency of our full-stack approach to sovereign AI infrastructure.”  Eric Rosa, Head of Engineering, Global AI.

‍

Global AI is the world’s first sovereign AI hyperscaler, providing dedicated AI infrastructure that combines the performance and operational sophistication traditionally associated with hyperscale cloud providers with the security, control, and isolation of single-tenant environments.

The company currently operates one of the largest NVIDIA GB300 NVL272 deployments in New York and is executing a disciplined expansion strategy targeting 100 megawatts of critical AI capacity by the end of 2026, 250 megawatts by the end of 2027, and one gigawatt by the end of 2029.

The NVIDIA Exemplar Clouds initiative recognizes NVIDIA Cloud Partners that achieve workload performance, security, and resiliency aligned to NVIDIA reference-architecture standards, giving developers and AI leaders greater confidence in performance consistency and total cost of ownership across cloud environments.

About Global AI

Global AI is the world’s first sovereign AI hyperscaler, purpose-built to design, build, power, and operate dedicated AI infrastructure for the world’s most critical AI workloads. The company provides nations and enterprises with single-tenant, isolated environments engineered for advanced AI training and inference workloads, enabling customers to maintain full control of their data, infrastructure, and operations while benefiting from hyperscale-class performance and operational expertise. Through its vertically integrated model spanning land, energy, construction, advanced liquid cooling, and GPU-dense compute, Global AI delivers secure, high-performance sovereign AI infrastructure at national scale.

‍