Jeff Denworth Co Founder
VAST Data and Hyperfusion have announced plans to integrate the VAST AI Operating System into modular AI infrastructure to accelerate the rollout of sovereign computing facilities. As a result of this collaboration, enterprises can deploy production computing clusters in weeks rather than waiting years for conventional builds. UAE-based Hyperfusion will use the software across its transportable data units.
Deploying modular AI infrastructure at Scale
Hardware procurement has outpaced facility availability across the global market. While organizations can secure processors quickly, securing sufficient power, cooling, and facility space remains difficult. Hyperfusion resolves this bottleneck by delivering factory-built modules directly to sites with available power connections.
The company packages an operational computing facility into a 45-foot ISO High Cube container known as Sentinel. Specifically, each unit includes a reference setup of 96 graphics processing units and up to three petabytes of storage. Meanwhile, larger facilities called HyperPod expand this setup to five megawatts and twenty petabytes to support complex artificial intelligence operations.
Hardware Specifications and Data Architecture
Under the new technical agreement, Hyperfusion integrates the VAST data platform as its preferred system. This integration allows distributed computing nodes to share a single global namespace across separate sites. Consequently, this modular AI infrastructure maintains uniform data storage parameters whether a customer runs a single container or ten interconnected units.
“Every government and enterprise we speak to wants the same three things: capacity now, data that never leaves their control, and a platform that grows without starting over.”
Quentin Reyes, Founder and CEO of Hyperfusion
Security Layers and Sovereign Control
Data sovereignty represents a core requirement for public sector and enterprise operators. Because Sentinel supports confidential computing as a security layer, organizations protect running models directly in active memory. In addition, systems can run completely air-gapped from external networks to ensure complete digital isolation.
Hyperfusion offers these units as customer-owned property or through fully managed operational service contracts. Furthermore, strict local controls guarantee that sensitive digital information remains strictly in-country. These measures protect critical assets while meeting regional data governance standards.
“The AI infrastructure conversation has been dominated by how many GPUs you can secure, when increasingly the gating factor is where you can actually put them. Power isn’t arriving in neat, hundred-megawatt blocks on the timetable AI demands.”
Jeff Denworth, Co-Founder at VAST Data
Market Outlook and Multi-Site Operations
Both companies are commencing joint commercial activities to target government, energy, and telecommunications customers. In addition, the partners plan to evaluate a unified control plane for managing distributed installations across varied geographical locations.
Demand for localized inference clusters continues to climb as computer vision and multimodal models expand. By combining transportable hardware with reliable data management, modular AI infrastructure allows operators to place intense computing workloads next to available energy sources without generating fragmented data silos.
