Vertiv Introduces Pre-Engineered Infrastructure for High-Density AI Data Centres

Last Updated 6 hours ago by Kenya Engineer

Vertiv has introduced a pre-engineered infrastructure platform designed to simplify the deployment of high-density artificial intelligence (AI) computing systems in existing buildings and new data centres, bringing power distribution, liquid cooling and monitoring into a factory-validated architecture.

Announced in Johannesburg, South Africa, on 29 October 2026, the Vertiv SmartIT solution targets enterprises, research institutions, colocation providers and hyperscale data centre operators seeking to accommodate increasingly demanding AI workloads without redesigning their entire facilities.

The platform integrates rack infrastructure, direct-to-chip liquid cooling, in-rack cooling distribution, direct current (DC) power conversion and distribution, and monitoring systems. Factory validation is intended to reduce the amount of integration required on site, leaving facility connections and commissioning as key stages of deployment.

The approach addresses a growing infrastructure challenge as AI computing systems place greater demands on electrical capacity, heat removal and coordination between IT equipment and building services. High-density computing racks require power and cooling systems designed to operate together, making infrastructure integration an important part of data centre planning.

Retrofitting existing buildings for AI workloads

A central feature of SmartIT is its intended use in both new facilities and existing buildings. Vertiv says the platform can be deployed in enterprise facilities, commercial buildings, colocation sites and hyperscale environments without requiring a complete redesign.

For existing facilities, the practical question is whether the available electrical supply, cooling capacity, floor loading, equipment access and supporting infrastructure can accommodate the proposed computing density. A pre-engineered rack solution may reduce integration work, but the building’s underlying capacity and the requirements of the selected AI hardware will still determine what can be installed safely and reliably.

The platform is designed to scale from a single rack to multi-rack installations. This allows operators to add capacity in stages using a common infrastructure architecture rather than developing a different arrangement for each expansion phase.

Vertiv also highlights integrated leak detection, automated response capabilities and continuous monitoring. These functions are particularly relevant to liquid-cooled systems, where cooling distribution, leak detection and operational controls need to work together to protect equipment and maintain service continuity.

Supporting next-generation AI computing

SmartIT is designed for AI infrastructure based on NVIDIA MGX and Open Compute Project (OCP) ORV3 architectures. Vertiv lists support for NVIDIA GB300 NVL72, Vera Rubin NVL72, DGX Rubin NVL8 and HGX Rubin NVL8 systems.

The solution was previously deployed with NVIDIA to support the NVIDIA DGX GB300 system. Vertiv says the platform is available globally.

Operational monitoring supports Redfish, Simple Network Management Protocol (SNMP) and Modbus, enabling integration with relevant equipment-management and monitoring systems. Vertiv Global Services provides deployment, commissioning and lifecycle support.

The company also positions SmartIT within a wider infrastructure offering. Vertiv SmartRun extends integration across a row of racks, while its rack power distribution units provide power distribution and monitoring. Vertiv Unify brings monitoring, analytics and controls together across power and thermal infrastructure, including leak detection and coordinated response functions.

Implications for data centre planning

As AI workloads expand, data centre projects must coordinate computing hardware, electrical infrastructure, cooling systems and operational monitoring from the design stage. Factory-integrated solutions offer one route to reducing the complexity of bringing these systems together, particularly where deployment schedules are tight or capacity is being added incrementally.

However, pre-engineered infrastructure does not remove the need for site-specific engineering. Operators must still establish that available power, heat rejection, redundancy, environmental conditions and operational procedures meet the demands of the intended workload.

For facilities considering AI upgrades, the key decision is not only which computing platform to install, but also how to deliver and distribute power, remove heat and maintain reliable operation as computing density increases.

Vertiv is a global provider of critical digital infrastructure for data centres, communication networks and other applications requiring continuous power and thermal management.