Rompower Energy Systems to Present Advanced AI Power Architectures at PCIM Seminars

Rompower Energy Systems' President, Dr. Ionel "Dan" Jitaru, will present advanced power architectures for AI applications at PCIM Europe 2026. The seminars will focus on ultra-high-efficiency DCX and LLC-based architectures, multi-leg magnetic structures, high-frequency operation, and GaN/SiC technologies for high power density and efficiency. The presentations will cover medium and low-power converters, soft-switching techniques, leakage-energy harvesting, and AI auxiliary power applications.

Rompower Energy Systems' President, Dr. Ionel "Dan" Jitaru, will present advanced power architectures tailored for artificial intelligence (AI) applications at PCIM Europe 2026. The seminars will explore cutting-edge technologies and design strategies aimed at improving efficiency and power density in AI infrastructure, addressing the growing demand for high-performance computing and energy-efficient systems.

A central focus of the presentations will be ultra-high-efficiency DCX and LLC-based architectures. These designs are critical for managing the complex power delivery needs of AI systems, particularly in high-voltage DC (HVDC) environments. DCX, or DC transformers, enable efficient bus conversion at high power levels, supporting deeper integration of power distribution within server racks and reducing conversion stages before power reaches the processor. LLC resonant converters, known for their soft-switching characteristics, are widely used in AI environments for their ability to maintain high efficiency across varying load conditions.

Dr. Jitaru will also discuss multi-leg magnetic structures and high-frequency operation, which are essential for achieving compact and efficient power conversion. These techniques help reduce thermal losses and improve system performance, particularly in applications where space and energy efficiency are critical. Additionally, the seminars will cover the use of gallium nitride (GaN) and silicon carbide (SiC) technologies, which offer significant advantages over traditional silicon-based semiconductors. GaN and SiC enable faster switching, reduced conduction losses, and higher power density, making them ideal for next-generation AI infrastructure.

The presentations will also address medium and low-power converter designs, emphasizing soft-switching techniques that minimize switching losses and improve overall system efficiency. These methods are particularly relevant in AI applications where frequent transient load changes require rapid response from power delivery systems. Dr. Jitaru will highlight the importance of leakage-energy harvesting, a technique that captures and reuses lost energy, further enhancing system efficiency.

In the context of AI auxiliary power applications, the seminars will explore how advanced power architectures can support the unique demands of AI accelerators and processors. These systems require precise voltage regulation and high transient response capabilities, which are increasingly being addressed through next-generation voltage regulator modules (VRMs) and trans-inductor voltage regulators (TLVRs). As AI workloads continue to evolve, the integration of these technologies will play a key role in enabling scalable, high-performance computing environments.

The growing adoption of AI is reshaping power infrastructure, with data centers transitioning into AI factories that require robust, energy-efficient power solutions. The use of 800 VDC power architectures, for instance, is gaining traction as a means to reduce conversion stages and improve efficiency in high-density computing environments. These trends underscore the importance of innovative power design in supporting the next phase of AI development.

Dr. Jitaru’s seminars will provide valuable insights into the technical and strategic considerations shaping the future of AI power systems. As AI workloads continue to scale, the ability to deliver reliable, high-efficiency power will remain a critical factor in determining the performance and sustainability of AI infrastructure.