Advanced gallium nitride and silicon carbide technologies aim to improve efficiency in next-generation AI factories.
Navitas Semiconductor is expanding its work with NVIDIA’s MGX ecosystem as demand grows for more powerful and energy-efficient AI infrastructure. The collaboration focuses on supporting next-generation AI data centers built around emerging 800VDC power architectures.
At COMPUTEX 2026 in Taipei, Navitas unveiled its latest 800V-to-6V DC-DC power delivery board (PDB), designed to address the rising power demands of AI servers and large-scale AI factories. The system is powered by the company’s GaNFast gallium nitride technology, which is aimed at improving efficiency while reducing the physical footprint of power conversion systems.
A key feature of the design is the elimination of the traditional 48V intermediate bus converter stage commonly used in servers. By simplifying the power delivery path, the solution reduces component complexity, improves reliability and creates additional space for computing hardware inside AI racks.
According to Navitas, the power delivery board can achieve peak efficiency of up to 97.5% while operating at a switching frequency of 1MHz. Its compact form factor allows it to be positioned close to GPU boards, helping improve power distribution and system responsiveness.
As AI models become larger and more compute-intensive, efficient power management has become a critical challenge for data center operators. Growing AI clusters require higher power density, better thermal performance and lower energy losses to maintain performance at scale.
Beyond its GaNFast portfolio, Navitas is also leveraging its GeneSiC silicon carbide technology to improve power conversion throughout AI infrastructure. These solutions are designed for applications such as solid-state transformers and high-power power supply units, enabling more efficient delivery of electricity from the grid to AI computing systems.
Through its gallium nitride and silicon carbide technologies, Navitas aims to help build the next generation of AI factories. The company believes advanced power solutions will be essential to improving efficiency, reliability and scalability as AI infrastructure continues to expand worldwide.
