TAIPEI, June 2 — Global energy technology leader Schneider Electric has unveiled an innovative 800-volt direct current (800VDC) power system designed to support NVIDIA’s next-generation graphics processing units (GPUs) and address the rising electricity demands of artificial intelligence (AI) data centres.
Known as the “sidecar”, the 800VDC power infrastructure relocates power conversion equipment outside the computing rack, freeing up valuable space for AI processors while reducing congestion within information technology (IT) infrastructure.
Schneider Electric Senior vice-president Himanshu Prasad said the solution was developed to help data centre operators manage rapidly rising power requirements as next-generation AI workloads become increasingly intensive.
By moving power conversion closer to the load, the sidecar improves energy efficiency, reduces power losses and enables more effective power delivery to high-performance AI systems.
The sidecar functions as a dedicated power rack positioned alongside AI server racks, helping data centres support the higher power densities required for next-generation AI computing.
Himanshu said advances in AI computing are driving a sharp increase in data centre power requirements as operators deploy increasingly powerful GPUs.
“As AI computing performance improves, the overall power levels required by data centres are also increasing.
“By increasing the voltage and reducing the current, you can reduce losses and make the system more efficient for future AI data centres,” he told a media briefing held in conjunction with Computex 2026 in Taipei today.
Computex 2026, a global technology and AIoT (artificial intelligence of things) trade exhibition, runs from June 2 to June 5 in Taipei, Taiwan.
The 800VDC sidecar, currently being showcased at the exhibition, was developed by Schneider Electric’s research centre in Taiwan.
Himanshu said traditional power systems used in data centres are becoming inadequate as AI workloads push rack power requirements beyond current limits, creating a need for more efficient power distribution solutions.
He said the transition to 800VDC is not based on a single technology but represents a phased evolution incorporating centralised power systems and advanced solid-state transformer technologies.
Beyond power conversion, he said the future of AI data centres would also depend on reliable protection systems, supply chain readiness, maintenance capabilities and operational expertise.
“It’s not just about the 800-volt architecture. It’s also about protection, supply chain, maintenance and service.
“We have to provide the complete solution for customers,” he said.
As global demand for AI infrastructure accelerates, Himanshu said next-generation power technologies would play a critical role in enabling larger and more energy-intensive AI factories and data centres.
