CRA Data Centres are Preparing for AI and Extreme Computational Performance
The rapid development of artificial intelligence and high-performance GPUs is fundamentally changing the requirements for data centre infrastructure. Increasing demands on computing power, energy density and efficient cooling are pushing technology standards and forcing infrastructure operators to look for new approaches to the design and operation of data centres. Therefore, the data centre services of České Radiokomunikace (CRA) are undergoing a gradual transformation that reflects the advent of AI and high-performance computing platforms. CRA is expanding its services and technological facilities to support modern AI workloads while ensuring high reliability, energy efficiency and long-term scalability of the infrastructure.
AI Development on CRA infrastructure
CRA is expanding its portfolio with AI-focused services that provide not only computing power based on GPU accelerators, but also consulting support for customers in the design and operation of AI solutions. This approach reflects the growing demand from companies for scalable and powerful platforms for model training, inference and big data.
AI workloads have specific infrastructure requirements, particularly an extreme density of computing power within a single rack, which leads to a dramatic increase in power consumption.
Limitations of Existing Infrastructure
CRA's existing data centres, including DC Lužice, have historically been designed for standard IT workloads. However, thanks to continuous upgrades, they are still able to operate technologies with power inputs of up to approximately 20 kW per rack. This allows you to cover a wide range of applications, including less demanding AI workloads or hybrid environments.
However, modern GPU servers for AI training and HPC (High Performance Computing) are now routinely well above this limit, often in the range of 30-80 kW per rack, and in some cases even more. This generates pressure to innovate data centre infrastructure.
Prague Gateway DC: Next-generation Infrastructure
CRA is responding to these technological trends by building a new data centre known as Prague Gateway DC, which is designed to be an infrastructure ready for the era of AI and extreme computing loads. Key technological innovations include:
1. Advanced Cooling (Liquid Cooling)
Traditional air cooling is becoming less and less effective at high power densities. Prague Gateway DC therefore counts on the implementation of liquid cooling, which significantly increases the ability to dissipate heat directly from components (e.g. GPUs). This approach:
- enables the operation of racks with many times higher power consumption,
- increases energy efficiency (lower PUE),
- reduces dependence on air conditioning systems.
2. Modern power architecture (800 V)
Increasing energy demands are leading to innovations in the field of electricity distribution. Prague Gateway DC takes into account trends such as the deployment of higher supply voltages (e.g. 800 V), which brings:
- lower losses in power transmission,
- more efficient power distribution to racks,
- better readiness for extreme loads on AI infrastructure.
3. Innovative Backup Solutions (Graphene Batteries)
Ensuring continuous availability is also crucial for AI services. Therefore, CRAs are considering using advanced storage technologies such as graphene batteries, which can offer:
- higher performance and faster response than traditional UPS systems
- longer service life,
- more compact dimensions
Hybrid Infrastructure Model
The CRA strategy is based on a combination of its existing data centres (e.g. DC Lužice) and the construction of new, technologically advanced locations. This allows:
- effective use of current investments,
- flexible migration of customers according to their needs,
- covering the entire spectrum of workloads from standard IT systems to demanding AI projects.
Key Conclusions
The evolution of CRA data centre services clearly reflects the global trend – the transition from a universal infrastructure to highly specialized platforms optimized for AI and HPC. The ability to handle extreme energy density, ensure efficient cooling, and innovate in power and backup are key success factors.
Thanks to the combination of modernized existing data centres and a new generation of infrastructure at Prague Gateway DC, CRA is ready to provide customers with a stable and powerful environment for the operation of modern digital services, including demanding AI projects and data-intensive applications, throughout the region.