The last decade has witnessed a profound transformation in global electricity consumption. Even under sluggish economic conditions, the compound annual growth rate of electricity demand over the past five years has approached 7%, driven by consumer electronics, cloud infrastructure, and above all, data centers. These facilities, fueled by the digital economy and the exponential uptake of artificial intelligence technologies, are reshaping the very landscape of electrical power distribution.

As workloads scale and single-rack power demand soars beyond 1 MW, the once-adequate supply frameworks are being tested to their limits. Copper usage, floor space allocation, and conversion efficiency have become decisive parameters in modern design. Against this backdrop, new supply methodologies are emerging to address inefficiencies inherent in legacy systems.

Traditional Low-Voltage Power Supply Systems

For decades, the backbone of data center infrastructure has been the 10 kV/0.4 kV alternating current distribution model. Power is stepped down through transformers and stabilized using uninterruptible power supplies (UPS) at both input and output stages. This arrangement guarantees supply quality and reliability, fulfilling the operational requirements of most small and mid-sized facilities.

The strengths of this design are evident: proven maturity, abundant equipment availability, and strong compatibility with existing servers and IT loads. Manufacturers offer high-quality transformers, switchgear, and cabinets at competitive costs, making the architecture a cost-effective and dependable option. Unsurprisingly, this system remains the prevailing choice for retrofits and medium-scale deployments.

Yet, its drawbacks are no less clear. The system suffers from multiple conversion layers, each introducing incremental inefficiencies. From 10 kV down to 0.4 kV and beyond, losses accumulate, driving up operational expenditures. For hyperscale centers where efficiency margins translate directly into millions in energy costs, the traditional architecture is increasingly viewed as a hindrance to scalability.

High-Voltage Direct Current (HVDC)

High-voltage direct current distribution has emerged as an alternative to traditional UPS-dominated frameworks. By removing the inverter stage, HVDC streamlines the conversion chain, trimming losses and improving system-wide efficiency. Batteries connect directly to the output bus, enabling instantaneous switchover in the event of an outage.

The benefits are substantial: elevated power density, reduced footprint compared to equivalent UPS setups, and lower copper demand due to smaller transmission currents. With voltages climbing from 240 V/336 V to 400 V and even 800 V, line losses decline further, making the solution increasingly appealing for high-density racks.

However, the inherent nature of DC introduces unique engineering challenges. Without a natural zero-crossing point, electrical arcs persist longer under fault conditions, demanding advanced protection devices with superior arc-extinguishing capabilities. Furthermore, HVDC is still in its maturation stage. Standards are fragmented, compatibility with server power supply units (PSUs) remains inconsistent, and integration costs are elevated. While pilot projects are underway, widespread adoption may still take several years.

"Panama Power" / Medium-Voltage Direct Supply

Often described as the next evolutionary step beyond HVDC, the so-called “Panama Power” or medium-voltage direct supply architecture eliminates layers of inefficiency by converting incoming 10 kV AC directly into 240 V/336 V DC. This streamlined approach minimizes conversion stages, curtails energy losses, and simplifies distribution hierarchies.

Efficiency rates exceeding 97% have been recorded, outperforming traditional AC-UPS configurations by a considerable margin. The architecture reduces space requirements by doing away with bulky UPS arrays, enabling higher rack density and improved spatial economics. The modular and redundant design further enhances resilience, ensuring that localized failures remain contained without cascading disruptions.

Despite these advantages, challenges remain. The supporting ecosystem is still embryonic. Most commercial servers are engineered for AC input, with DC-compatible PSUs occupying only a niche market. This limits flexibility and inflates procurement costs. Voltage standardization also lags behind, creating uncertainty for large-scale deployment. Nonetheless, as hyperscale data centers chase efficiency at unprecedented scales, medium-voltage direct supply is increasingly viewed as a blueprint for the future.

Conclusion

The evolution of low-voltage supply systems in data centers reflects the industry’s larger trajectory: a relentless pursuit of efficiency, density, and resilience. Traditional AC frameworks retain their relevance in smaller and legacy contexts. HVDC offers an intermediate path, balancing familiarity with incremental performance gains. Medium-voltage direct supply, though nascent, signals the long-term shift toward ultra-efficient, space-saving, and simplified infrastructures.

The rise of artificial intelligence and data-driven industries ensures that power architectures will remain at the heart of innovation. For operators and designers, the challenge is not merely choosing a system that works today, but preparing for the demands of tomorrow’s computational landscape.