Vertiv, NVIDIA & iGenius to deploy Colosseum AI supercomputer

Vertiv has announced a collaboration with NVIDIA and iGenius to deploy Colosseum, an NVIDIA DGX AI supercomputer using NVIDIA Grace Blackwell Superchips, in Italy in 2025.

According to Vertiv, Colosseum is designed to meet the needs of highly regulated industries including finance, healthcare, and public administration, with a focus on computational power, energy efficiency, data sovereignty, and data security.

The deployment of Colosseum marks the next stage in the ongoing partnership between Vertiv and NVIDIA, and it will be strategically located in southern Italy to address the requirements of local governments.

Uljan Sharka, Chief Executive Officer of iGenius, said, "Harnessing the power of NVIDIA's cutting-edge accelerated computing and Vertiv's innovative infrastructure expertise, Colosseum stands as a testament to the transformative potential of sovereign AI. We're demonstrating how modular systems and software-specific infrastructure enable a new era of mission-critical AI."

Colosseum is based on the NVIDIA DGX SuperPOD architecture, combining Vertiv's infrastructure management, NVIDIA accelerated computing and the NVIDIA Omniverse Blueprint for AI factory design and operations. The AI supercomputer will leverage Vertiv's 360AI reference architecture infrastructure, designed specifically for NVIDIA GB200 NVL72, which was co-developed with NVIDIA and released in late 2024.

The system's modular and scalable design supports iGenius in deploying one of the fastest sovereign AI supercomputers at hyperscale. Vertiv has also expanded its reference design library for AI data centres through its AI Hub, offering a power and cooling system for the NVIDIA GB300 NVL72. This approach allows customers to anticipate infrastructure needs ahead of the latest GPU generation deployments with ready-made designs for data centre scale-out.

Karsten Winther, President of Vertiv, EMEA, commented, "The unit of compute is no longer the chip - it's the system, the AI Factory. Through our collaboration with NVIDIA and visionary AI player iGenius, we are proving the efficiency and system-level maturity of delivering the data centre as a unit of compute, unlocking rapid adoption of AI-native power and cooling infrastructure as a catalyst for AI at scale."

The project also involves the use of NVIDIA Omniverse technologies for physically-accurate digital twins of Colosseum. This enables real-time collaboration between Vertiv, iGenius and NVIDIA, helping accelerate decisions and reduce the design-to-deployment timeline. Digital twin simulation reduces simulation times from several months to hours, and Vertiv's manufacturing and integration processes can cut deployment time by as much as 50% compared to traditional builds.

Charlie Boyle, Vice President of DGX Platforms at NVIDIA, said, "AI is reshaping the data centre landscape, demanding new levels of scale, efficiency and adaptability for global AI factories. With physically-based digital twins enabled by NVIDIA Omniverse technologies and Vertiv's modular design for the iGenius DGX SuperPOD data centre, Colosseum sets a new standard for building supercomputers for the era of AI."

The design of Colosseum was validated by a collaborative 3D design process, which included predictive modelling of thermal loads, electrical flows, and site layouts. The centre's racks, initially liquid cooled at 132kW each, will be scalable for future technologies.

The modular data centre solution provided by Vertiv encompasses design, manufacturing, installation, and commissioning, as well as ongoing management, monitoring, service, and maintenance. The building incorporates prefabricated white space within and a modular grey space outside for scalability and efficiency in deployment.

Colosseum will use NVIDIA Mission Control for its orchestration along with Vertiv Unify to simplify and synchronise building management. Vertiv Unify provides real-time orchestration across power, cooling, and compute, synchronisation via digital twins for closed-loop optimisation, and supports autonomous decision-making processes.

The integration of NVIDIA Omniverse technologies with Vertiv Unify will allow for real-time updates between the physical infrastructure and digital models, supporting predictive maintenance and scenario testing to minimise operational risks.

Colosseum is intended to serve as a template for scalable and repeatable sovereign AI factories, combining high-density cloud infrastructure, controlled data management, and modular deployment. The initiative will be extended globally, with ongoing alignment between iGenius, Vertiv, and NVIDIA to support future platforms, including DGX GB300 systems.