Infineon Technologies AG, a global leader in semiconductor solutions for power systems and IoT, has announced its entry into NVIDIA’s MGX AI Factory ecosystem to support the development of advanced power delivery systems for next-generation AI data centers. Through this collaboration, Infineon will contribute its power management technologies to NVIDIA’s MGX architecture and emerging 800 VDC power infrastructure. The initiative is designed to help AI data centers achieve greater computing density, improved energy efficiency, and scalable infrastructure for increasingly complex artificial intelligence workloads.
The NVIDIA MGX architecture is an open and modular platform built to accelerate AI factory deployments for the agentic AI era. By supporting 800 VDC-compatible power racks, the architecture allows existing AI infrastructure to increase computing performance while preparing for future high-density workloads without requiring complete infrastructure replacement.
Commenting on the collaboration, Adam White, Division President, Power & Sensor Systems at Infineon, said, “As AI models become larger and more computationally intensive, data centers are under pressure to deliver significantly higher performance within existing power, cooling, and physical constraints. Our work with NVIDIA focuses on redefining power delivery from the grid to the processor core to improve efficiency and support the next generation of AI infrastructure.”
Infineon’s expertise spans multiple semiconductor technologies, including silicon, silicon carbide, and gallium nitride, enabling efficient power conversion across the entire data center ecosystem. The company’s gallium nitride technology supports switching frequencies close to 1 MHz, enabling highly compact and efficient bus converters. In addition, its proprietary silicon carbide JFET technology and specialised control ICs support protection and hot-swap functionality for native 800 V server systems.
The collaboration aims to streamline the complete power conversion process from high-voltage input down to intermediate and core voltages used in AI computing systems. By reducing conversion stages and delivering DC power closer to server racks, the architecture improves overall efficiency, simplifies infrastructure design, and supports higher-density AI deployments.
NVIDIA’s 800 VDC-compatible MGX power racks are intended to provide a practical upgrade path for enterprises looking to scale accelerated computing capacity while preserving existing infrastructure investments. The platform also enables hybrid power architectures that support future AI workloads and evolving data center requirements.
Infineon currently employs approximately 57,000 people worldwide and reported revenue of around €14.7 billion during the 2025 fiscal year. The company continues to focus on advancing digitalisation and decarbonisation through innovative semiconductor technologies and energy-efficient solutions.
