IN Brief:
- Navitas will demonstrate GaN and SiC power technologies at PCIM Europe 2026 in Nuremberg.
- The demonstrations include 800V-to-6V and 800V-to-50V converter platforms for AI data centre power architectures.
- Data centre, grid, and industrial electrification systems are converging around higher efficiency, higher voltage distribution, and denser conversion stages.
Navitas Semiconductor will showcase GaN and SiC power technologies at PCIM Europe 2026, with demonstrations aimed at AI data centres, energy infrastructure, grid conversion, and industrial electrification.
The company’s PCIM line-up includes GaNFast FETs, GaNSafe and GaNSlim power devices, bidirectional GaN ICs, GeneSiC silicon carbide devices, and SiCPAK power modules. The product range spans low-resistance 100V GaN devices through to 3300V and 2300V SiC modules for higher-voltage conversion systems.
For AI data centres, Navitas will show a 20kW 800V-to-6V GaN-based power delivery board targeting 97.5% peak efficiency. The design removes the traditional 48V intermediate bus converter stage, reducing the number of conversion steps between high-voltage distribution and processor-level power delivery.
A second AI data centre platform provides 10kW 800V-to-50V conversion, with 98.5% peak efficiency and power density of 2.1kW per cubic inch. The full-brick design uses 650V and 100V GaNFast devices in a three-level half-bridge architecture with synchronous rectification, supporting 800VDC and ±400V power architectures.
For grid and energy applications, the PCIM programme includes solid-state transformer demonstrators using Navitas GeneSiC ultra-high-voltage and high-voltage technologies. The company will show an EPFL-developed full SST cell that integrates the primary converter, transformer, and secondary conversion stage, as well as a 50kVA bidirectional active front-end dual-active-bridge SST solution based on 3300V SiCPAK MOSFET modules.
The same booth will also include industrial electrification and motor-control inverter solutions using GaNSense motor-drive ICs with integrated power stage, current sensing, voltage sensing, and temperature protection. Auxiliary power supplies, solar microinverter hardware, bidirectional GaN switch evaluation boards, and partner-developed high-power DC/DC conversion will also be included.
Power architecture is being reshaped by AI infrastructure as rack power levels rise and conventional distribution schemes come under pressure. Higher-voltage distribution can reduce copper losses and improve system efficiency, but it also raises the demands placed on isolation, protection, magnetics, control, and validation. GaN and SiC devices are central to that shift because they support faster switching, lower losses, and smaller passive components when the surrounding design is engineered correctly.
Testing and production capability are moving alongside the device technology. Rohde & Schwarz’s expansion of power electronics testing and Cohu’s GaN power device test orders for AI data centre applications both point to a market where wide-bandgap power design is becoming a system-level discipline rather than a component substitution exercise.
Grid conversion is travelling along a parallel path. Renewable generation, storage, EV charging, and industrial electrification are creating more variable and bidirectional power flows across distribution networks. Solid-state transformers and high-voltage SiC conversion stages are being developed to improve controllability and efficiency in systems that must bridge medium-voltage infrastructure and lower-voltage DC architectures.
Navitas’ PCIM line-up brings those strands together. AI data centres, grid infrastructure, and industrial power systems differ in scale and operating duty, but they are all pushing toward higher voltage, higher density, and tighter control. Wide-bandgap design is moving from device selection into complete conversion architecture, where efficiency gains depend on packaging, magnetics, thermal paths, control firmware, and validation as much as semiconductor performance.
