iGenius is launching one of the world’s largest AI data centres, powered by Vertiv infrastructure, NVIDIA Accelerated Computing and Omniverse

• Supports enterprise-wide deployment of Sovereign AI Compute.
• Combines exceptional computing power, energy efficiency and data sovereignty, and meets stringent data security requirements.

Vertiv, a global provider of mission-critical digital infrastructure, announces a groundbreaking collaboration with NVIDIA and renowned AI pioneer iGenius. The goal is to deliver Colosseum, one of the world’s largest AI supercomputers based on NVIDIA DGX with NVIDIA Grace Blackwell superprocessors. Colosseum will redefine the digital landscape as a unique sovereign AI data centre for regulated workloads and is scheduled to go live in Italy in 2025.

Designed to meet the needs of highly regulated industries such as finance, healthcare and public administration, Colosseum will embody a fusion of transformative computing power, energy efficiency and data sovereignty, while meeting the most stringent data security requirements.

Colosseum is based on an NVIDIA DGX SuperPOD and is the latest development in a long-standing collaboration between Vertiv and NVIDIA. The project is strategically located in southern Italy to meet regional regulatory requirements and marks a significant milestone in the European AI landscape.

‘Colosseum demonstrates the transformative potential of Sovereign AI by leveraging the power of NVIDIA’s state-of-the-art accelerated computing technology and Vertiv’s innovative infrastructure expertise,’ said Uljan Sharka, CEO of iGenius. ’We are demonstrating how modular systems and a software-based infrastructure enable a new era of business-critical AI.’

Modular design. Built for efficiency.

Colosseum combines Vertiv’s expertise in infrastructure management, NVIDIA Accelerated Computing and the NVIDIA Omniverse Blueprint for designing and operating AI factories. The project utilises Vertiv’s 360AI reference architecture platform for data centre power and cooling. It was developed for NVIDIA GB200 NVL72 infrastructure, which was designed in collaboration with NVIDIA and launched at the end of 2024. The modular and scalable system enables iGenius to deliver one of the fastest hyperscale AI supercomputers and one of the largest supercomputers to support sovereign AI.

In addition, Vertiv has expanded its reference design library in the AI Hub with the jointly developed Data Centre Power and Cooling Design for the NVIDIA GB300 NVL72 system. Because Vertiv is always one GPU generation ahead, customers can design their infrastructure before the chips are available, with finished designs that anticipate increasing power density in the rack and replicable models for large-scale AI factories.

‘The unit of measurement for computing power today is no longer the chip – it’s the system, the AI factory,’ said Karsten Winther, President, Vertiv, EMEA. ’Together with NVIDIA and visionary AI specialist iGenius, we are demonstrating the level of efficiency and system maturity that can be achieved when the data centre itself is designed and deployed as a single unit. This paves the way for the rapid adoption of AI-native power and cooling infrastructures as a catalyst for AI at scale.’

Simulate with NVIDIA Omniverse. Deliver quickly.

‘AI is reshaping the data centre landscape and requires new dimensions of scalability, efficiency and adaptability for global AI factories,’ said Charlie Boyle, Vice President of DGX Platforms at NVIDIA. ’With physically accurate digital twins enabled by NVIDIA Omniverse technologies and Vertiv’s modular design for

the iGenius DGX SuperPOD Data Centre, Colosseum is setting a new standard for building supercomputers in the AI era.’

Colosseum was developed as a physically accurate digital twin using NVIDIA Omniverse technologies. This approach enables real-time collaboration between Vertiv, iGenius and NVIDIA to accelerate system-wide decisions and shorten the cycle from design to deployment. The Omniverse Blueprint also enables real-time simulations, allowing engineers to test and refine designs immediately rather than waiting for lengthy simulation processes, reducing simulation times from months to hours. Vertiv’s manufacturing and factory integration processes reduce deployment time by up to 50 percent compared to traditional data centre projects.

This collaborative 3D design process validates the entire infrastructure stack and enables predictive modelling of heat load, power flow and site layout – from 132 kW liquid-cooled racks to modular power supply systems – before a single module is built.

Innovatively designed. Unified by software.

The AI-enabled, prefabricated modular data centre solution is designed, manufactured, delivered, installed and commissioned by Vertiv. It includes power, cooling, management, monitoring, service and maintenance offerings, with power and cooling capacity initially supporting up to 132 kW/rack and scalable for future designs as needed. The building envelope integrates prefabricated white space, while a fully modular grey space is provided on the outside. This approach offers exceptional scalability and energy efficiency, transforming the way data centres are built and deployed.

Colosseum will use the NVIDIA Mission Control system to operate and orchestrate the data centre, as well as Vertiv Unify to simplify and synchronise building management for AI factories.

Vertiv Unify offers:

· Real-time orchestration of energy, cooling and data processing, as well as digital twin synchronisation for closed-loop optimisation

· AI-enabled features that support autonomous decision-making

By integrating NVIDIA Omniverse technologies, Vertiv Unify enables real-time updates between physical systems and digital models, enabling predictive maintenance, what-if simulations and scenario testing before operational risks arise.

The global blueprint for AI factories

Colosseum is more than just a data centre; the project is a blueprint for scalable, standardised and sovereign AI factories. By combining cloud-scale density, local data control and modular deployment, the data centre demonstrates the next phase of AI, where inference must be secure, fast, compliant and distributed.

Colosseum is not a one-off project – it is a reference. With Colosseum, iGenius is building a blueprint that can be replicated on a global scale. Vertiv and NVIDIA have already agreed on future platform support, including DGX GB300 systems and beyond. This means that the future of sovereign AI is no longer a theory – it is already becoming a reality.