Data centers of the future: powered by superconducting cables for scalable and sustainable growth

Powering gigawatt-scale data centers generally relies on two voltage levels:

• Cables located outside the buildings connect the data halls to the transmission or distribution grid. They power Data Centers from the Primary and secondary HV Substation with HV and/or MV cables.
• Inside the buildings, Data Halls racks and IT equipment are connected at low-voltage level with distribution cables.

Each voltage level, comes with its own challenges in meeting the enormous energy demands of data centers such as AI Data Centers.

At the medium-voltage level, the total power of a data center is distributed across its various data halls, where the servers are hosted. To give a sense of scale: a next-generation data center may require between 100 MW and 400 MW per data hall, depending on its size and design choices. Supplying such power requires a very large number of parallel cables. Since these cables are generally buried underground, this leads to several issues:

• a large physical footprint,
• very high civil engineering costs,
• and electrical losses (Joule losses), which reduce efficiency and heat the soil.

Example:
Supplying a single 300 MW data hall requires 36 large-section cables (600 mm²) at 33 kV. A 1.8 GW data center with six 300 MW data halls would therefore require 216 buried cables—a massive, complex, and extremely costly infrastructure.

At the low-voltage level, once close to the data hall, a transformer steps down medium voltage to low voltage to feed the IT equipment directly. Delivering 100 MW to 400 MW at 480 V or 600 V results in extremely high currents, ranging from 6 kA to 10 kA.

Such currents require a very large number of low-voltage cables, installed inside the data hall in busways. This solution brings several constraints:

• a significant footprint (space occupation),
• a complex and costly cable installation architecture,
• strict electromagnetic compatibility requirements, both between cables in the same busway and with nearby equipment,
• and thermal management: conventional copper cabling generates significant heat, driving increased demand for HVAC systems and contributing to operational inefficiencies.

On top of this, Joule losses alone can reduce the overall efficiency of the data center by 5 to 10%. Today, some solutions even consider cooling low-voltage cables to limit these losses and improve efficiency. But this already highlights the limitations of conventional cabling solutions.

Superconductivity is not limited to energy transport: it also helps protect electrical networks through Superconducting Fault Current Limiters (SFCLs). Designed to absorb high fault currents in medium- and low-voltage grids, particularly in data centers, SFCLs significantly reduce—or even eliminate—overcurrents. This ensures service continuity, a critical requirement for such infrastructures. Their use also relieves circuit breakers and power electronics from excessive stress, simplifying the entire electrical architecture, from transformers to IT equipment.

The fundamental principle is as follows: in normal operation, an SFCL has zero resistance and does not affect the grid. However, when a fault drives the current above its nominal limit, the superconductor immediately transitions into a resistive state. This resistance is temporary, appearing only during the fault, and serves to sharply reduce the fault current. For example, in a data center grid where medium-voltage fault currents range from 20 kA to over 50 kA at peak, the presence of an SFCL can cut these values in half—or even more. This limits or eliminates fault while relieving stress on all electrical components within the infrastructure. As a result, the equipment’s lifespan is extended, maintenance costs are reduced, and most importantly, service continuity is maintained in a fully autonomous way.

Resilience through advanced protection

For data center operators, SFCLs offer protection capabilities that go beyond conventional systems:

• Enhanced protection of critical infrastructure including transformers, switchgear, and power distribution units
• Reduced risk of cascading failures that could affect multiple facility zones
• Improved power quality through automatic fault current management
• Lower maintenance requirements due to reduced mechanical stress on protective equipment

As digital demand grows, superconducting technology could transform data centers from energy-intensive liabilities into efficient, resilient participants in smart grids.

The compact, modular nature of superconducting systems allows data centers to locate closer to urban populations, reducing network latency while minimizing transmission losses from remote facilities. This supports the distributed infrastructure model essential for both digital services and grid stability. And as the digital economy continues its exponential growth, the efficiency gains from superconducting technology become essential for sustainable development

Companies like Nexans, leading in HTS cable systems and SFCLs, are driving this infrastructure revolution through R&D, manufacturing, and deployment, paving the way for next-generation data centers and a sustainable digital future.