Global AI Secures $441 Million for Sovereign Data Centers
Sovereign AI provider Global AI has raised $441 million in its first debt round to expand its isolated data centers, highlighting the growing demand for highly secure, private hardware.
Global AI, a startup specializing in single-tenant, air-gapped data centers, has secured a $441 million senior secured credit facility. Led and arranged by J.P. Morgan, this marks the company's first-ever debt raise. The funding comes alongside a broader surge in infrastructure financing, including a recent $900 million credit facility for GPU operator Nscale. Global AI plans to use the capital to scale its physical footprint to meet what CEO Sami Issa described as "exceptional demand" for isolated AI environments. The company currently boasts $6.2 billion in contracted revenues, with $1 billion already delivered to clients.
Founded in 2024, Global AI operates as a sovereign hyperscaler, designing physically isolated data centers for governments and enterprises that require strict data sovereignty. The company's hardware stack is built entirely on Nvidia technology. At its facility in Endicott, New York, Global AI currently runs GB200 NVL72 clusters. It recently completed a GB300 NVL72 deployment, which it claims is the largest of its kind in New York State. The firm is actively deploying 7,000 GB300 systems and plans to integrate Nvidia's next-generation Vera Rubin NVL72 racks across its U.S. facilities.
The newly secured debt will fuel a massive expansion of the company's power capacity. The Endicott site will scale from an initial 16 megawatts of critical IT load to 100 megawatts by the end of 2026. Global AI targets 250 megawatts in 2027 and aims for 1 gigawatt of total capacity by 2029. This expansion is supported by strategic partnerships, including a joint venture with HUMAIN, a Saudi Arabian Public Investment Fund-backed entity, to build large-scale U.S. data centers. Nvidia has also participated in Global AI's prior equity fundraising.
For AI practitioners and enterprise developers, this expansion offers a viable alternative to public cloud environments. By providing dedicated, air-gapped infrastructure, Global AI allows organizations to train and run massive models without exposing sensitive proprietary data or intellectual property to shared networks. The rapid deployment of cutting-edge hardware like the GB300 and Vera Rubin architectures ensures that highly regulated industries can access top-tier compute speeds while maintaining absolute jurisdiction over their workloads.
This is our own summary of reporting by Unite.AI
