In a strategic collaboration announced on October 13, 2025, Nvidia and Schneider Electric are teaming up to develop 800-volt direct current (VDC) power systems tailored for the escalating demands of AI data centers. This partnership aims to support Nvidia’s next-generation GPUs by enabling racks with power capacities up to 1.2 megawatts (MW), addressing the bottlenecks in traditional power infrastructures amid the AI boom. As data centers evolve into “AI factories,” this move highlights the industry’s shift toward high-voltage, efficient power delivery to handle the computational intensity of advanced AI models.
Nvidia, the Santa Clara-based chip giant founded in 1993, has dominated the AI hardware landscape with its GPUs powering everything from generative AI to supercomputing.
Under CEO Jensen Huang, the company has expanded beyond graphics into data center solutions, with its Blackwell platform representing a 3.4x density increase over predecessors like Hopper. Schneider Electric, a French multinational established in 1836, specializes in energy management and automation, with a strong focus on sustainable data center infrastructure. Led by CEO Peter Herweck, Schneider has been pivotal in digital transformation, reporting €35.9 billion in revenue for 2024. The duo’s alliance builds on earlier ties, including a June 2025 partnership for AI data center designs in Europe.
The core of the partnership is the development of an 800 VDC sidecar—a modular power unit capable of powering ultra-high-density server racks. This technology shifts from conventional 415- or 480-volt alternating current (VAC) three-phase systems to a high-voltage DC architecture, eliminating redundant conversions and streamlining power delivery. Key features include centralized AC-to-DC conversion at the facility level, followed by distribution via a three-wire system (positive, return, protective earth) to racks. Within the racks, late-stage converters like Nvidia’s Kyber step down to 12 VDC for GPUs. Integrated energy storage addresses power volatility: short-duration capacitors handle millisecond spikes, while battery systems manage longer ramps.
Benefits are multifaceted. The 800 VDC setup boosts end-to-end efficiency by up to 5%, reduces copper usage by 45%, and cuts transmission losses, leading to lower cooling needs and operational costs. It supports scalability for 1 MW+ racks, crucial as AI workloads drive rack densities from 100 kW to over 1 MW by 2027. Maintenance costs could drop by 70% due to fewer components and failures, potentially slashing total cost of ownership (TCO) by 30%. Environmentally, it promotes sustainability by minimizing energy waste, aligning with global net-zero goals. Schneider’s involvement ensures compatibility with existing infrastructures, while Nvidia’s ecosystem includes over 20 partners like ABB, Eaton, and Siemens for components and standards development through the Open Compute Project (OCP).
This collaboration extends beyond hardware. Schneider is committed to releasing power conversion and distribution products compliant with Nvidia’s 800 VDC architecture, fostering an interoperable ecosystem. Nvidia has published a whitepaper on the architecture and plans to present details at the 2025 OCP Global Summit. The timeline envisions initial deployments in 2026-2027, leveraging ecosystems from electric vehicles (EVs) and solar industries for rapid adoption. Executives emphasize innovation: Schneider’s Philippe Diez noted, “We are excited to collaborate with NVIDIA to bring forth power solutions that will enable the next wave of AI advancements.” Nvidia’s Brian Catlett highlighted the need for “a robust ecosystem to deliver efficient, scalable power.”
The implications ripple across the AI sector. With data center spending projected to surge from $39.49 billion in 2025 to $124.70 billion by 2030, efficient power systems are critical to avoid grid strains and blackouts. This partnership could accelerate AI infrastructure buildouts, benefiting hyperscalers like Microsoft and Google, which rely on Nvidia GPUs. However, challenges include safety standards for high-voltage DC, regulatory approvals, and integration with legacy systems. Critics note potential supply chain dependencies on rare materials, though reduced copper needs mitigate this.
Looking ahead, the alliance positions Nvidia and Schneider as leaders in the “gigawatt AI factories” era. As AI demands trillions in infrastructure investment, 800 VDC could become the standard, enabling denser, greener data centers. Complementary efforts, like Schneider’s liquid cooling solutions for AI, further enhance performance.
If successful, this could democratize high-performance AI, driving economic growth while curbing energy consumption. As Huang often says, AI is “the new electricity”—and with partners like Schneider, Nvidia is wiring the future.
Leave a Reply