For years, the idea of direct current (DC) power distribution has tempted the data center industry.

The appeal has always been clear: a simpler power flow and less equipment. Today, alternating current (AC) comes into a data center, gets turned into DC to charge the batteries, then back into AC to move around the facility, only to flip to DC at the server rack, to feed computers the power they need. That loop introduces losses and complexity at every step.

Throughout the 2010s, advocates made serious attempts to push for adoption. Major players like Intel, ABB, and IBM were involved in proof-of-concept deployments, and operators like Sakura Internet experimented with DC-powered campuses.

Yet the idea never took off.

Why it didn’t work, until now

The reasons for that failure were practical. AC gear was ubiquitous: standardized, available, and affordable. Density pressures hadn’t yet arrived, so computer racks could still fit server-level conversion equipment.

And, perhaps most crucially, the physics of copper limited designers’ options. At low voltages, copper losses balloon. At higher voltages, the copper needs to shrink, but new protection and arc-interruption challenges can endanger technicians. That left the industry stuck between tradeoffs: overbuilt copper systems that were massive and costly, or high-voltage DC architectures that strained safety and code compliance.

But the pressure is building

Today, the industry is under new pressures. Power delivery is a bottleneck thanks to AI demand, and racks once drawing 20 kilowatts now push above 200kW. That surge makes it physically difficult to fit all the needed conversion gear within the rack. Plus, power losses at each conversion stage now mean millions in wasted energy costs.

Studies are also clear on the competitive advantage of direct current. Nokia has demonstrated 90 percent efficiency for 400 VDC versus 77–84 percent for standard AC distribution. A 2018 IEEE review concluded that DC topologies are not only more efficient but also more reliable. And NVIDIA’s 2025 HVDC paper effectively demanded that the industry solve DC distribution at scale.

The industry’s attempted comeback

Under this pressure, new initiatives are taking shape. Vendors and hyperscalers are pushing for DC-compatible designs by 2028. These efforts include:

• 800 VDC campus designs to cut copper mass (pioneered by NVIDIA and Edge Cloud Link)
• 400 VDC telecom-proven platforms extended for data centers (e.g., Vertiv NetSure HVT)
• Ongoing standards work through Equinix–NUS collaborations, the EMerge Alliance, and NEMA task forces.

The direction of travel is becoming clearer—but the tradeoffs remain. Scaling DC means choosing between complexity, cost, and safety. Until now, no solution has fully resolved that triangle.

The missing enabler: High-temperature superconductors

That may have now changed.

High-temperature superconductors (HTS) offer a way to deliver bulk DC power at low voltages, without the copper bloat or electricity losses. When cooled, HTS wire can carry more than 250 times the current of copper by cross-sectional area, with no resistive heat loss. That shifts the equation that blocked DC designs for so long.

HTS products enable bulk DC distribution at low, safer voltages across a campus without the copper bloat. This reframes DC from a choice between massive conductors or hazardous voltages into a path that offers both density and safety. As a bonus, HTS systems can also inherently limit fault current, addressing a top safety concern at the source.

Superconductors are deployment ready

This isn’t hypothetical. HTS cables have been in utility-grade operation for over 20 years, with successful deployments in the United States, Japan, Germany, and France. In fact, great analogs for data center use are already public: Nexans operates two 5.3 MW, 1.5 kV DC HTS cables powering Paris’ Montparnasse train station. Sakura Internet demonstrated a 10 kV, 100MW HTS Direct Current link.

Vendors are moving too: VEIR with recent backing from Microsoft and National Grid, while other designers like Vision Electric Superconductors offer busbars for data centers. OEMs like LS Cable recently announced utility collaboration for a hyperscale project.

Momentum now spans the HTS industry, with 20 years of proof and a focus on data center needs.

What should happen next?

We don’t need to compromise safety or efficiency to reach direct current. HTS delivers both.

While data centers are adopting individual HTS products now, realizing this future will take more than a single vendor. It calls for a coalition: operators, OEMs, system designers, cablemakers, and standards bodies. Together, we can prove this model in a real facility: from on-site generation without voltage step-ups, through simplified no-loss campus cabling, to low-voltage busbars powering racks.

A national demonstration site would validate the architecture and help secure US leadership in critical infrastructure.

DC never stopped being attractive. It just lacked the right conductor. The missing enabler is finally on the table with HTS.