Wise Integration and Navitas Semiconductor have partnered to combine digital power control with gallium-nitride (GaN) and silicon-carbide (SiC) technologies. This collaboration aims to deliver highly efficient, next-generation power solutions tailored for the demanding energy requirements of AI data centers.
Overview of the Wise and Navitas Collaboration
Wise Integration and Navitas Semiconductor have announced a strategic collaboration, agreed upon on Sept. 25, 2026, to merge advanced digital power control with next-generation wide-bandgap semiconductors. This partnership is specifically engineered to address the escalating energy requirements of artificial intelligence computing and modern data-center power systems. The joint effort will primarily focus on developing AI data-center power supply units (PSUs) and other high-voltage, high-power conversion topologies. By bridging the gap between the electrical grid and compute silicon, the companies aim to deliver highly efficient power solutions tailored for the demanding environments of next-generation AI infrastructure.
Technical Details: Digital Control Meets Wide-Bandgap Semiconductors
The technical foundation of this collaboration relies on the synergy between distributed digital control and advanced semiconductor materials. Wise Integration brings WiseWare to the table, a sophisticated distributed digital-control technology designed specifically for power converters. Complementing this, Navitas Semiconductor contributes its industry-leading GaNFast gallium-nitride and GeneSiC silicon-carbide devices. The integration of these technologies is expected to significantly improve conversion efficiency, power density, and real-time control capabilities.
Wide-bandgap devices possess unique physical properties that allow them to switch at substantially higher frequencies and tolerate much higher voltages compared to traditional silicon. This capability is crucial because it potentially reduces the physical size of passive components while simultaneously increasing overall power density. Furthermore, the embedded digital control allows the system to adjust power-conversion behavior in real time, optimizing performance dynamically under the highly variable loads characteristic of AI workloads.
Architectural Shift: The Grid-to-Chip Semiconductor Problem
The partnership underscores a profound architectural shift within the data center industry. As AI facilities transition toward megawatt-scale racks and higher-voltage distribution paradigms, the electrical path between utility power and GPUs is increasingly becoming a specialized semiconductor problem. Navitas is already actively developing GaN and SiC technology to support emerging 800 VDC AI infrastructure architectures.
A prime example of this technological trajectory is Navitas's previously demonstrated 800 V-to-6 V power board. This design utilizes 650 V GaN devices and is explicitly engineered to eliminate the traditional 48 V intermediate conversion stage typically found inside server trays. By removing this intermediate stage, the architecture reduces conversion losses and simplifies the power delivery network, highlighting how the grid-to-chip transition relies heavily on specialized power semiconductors, digital controllers, and high-density PSUs.
Market Context: Rising Power Density and Expanding Applications
The primary catalyst for this technological convergence is the relentless increase in rack power density driven by advanced AI accelerators. These next-generation processors place immense pressure on traditional silicon conversion stages, which are struggling to meet the thermal and efficiency demands of modern AI clusters. Wide-bandgap solutions provide a viable pathway to overcome these physical limitations.
While AI data centers represent the primary target market, the applications for this combined digital and wide-bandgap technology extend far beyond hyperscale facilities. Wise and Navitas plan to explore these characteristics across a broader spectrum of high-performance computing. Target applications include AI PCs, high-end workstations, gaming desktops, and various industrial computing systems, all of which require compact, highly efficient, and dynamically controllable power delivery.
Industry Impact and Broader Ecosystem Trends
The Wise and Navitas collaboration is not an isolated event but rather a reflection of a broader industry-wide transformation in power and thermal management. The sector is witnessing a concerted shift toward higher-density power delivery, paralleling advancements in cooling and optical interconnects. Recent industry developments include ABB’s 800 VDC AI power system, Vertiv’s 2.3 MW AI cooling infrastructure, and Huawei’s Atlas 960E optical-power architecture.
Furthermore, the semiconductor supply chain is adapting to these new paradigms. Companies like onsemi are actively embedding power devices directly into silicon to achieve denser AI racks, while Enphase is manufacturing 800 VDC AI power modules. The consensus across the ecosystem is clear: the primary constraint limiting AI expansion is no longer just the delivery of compute performance, but the ability to efficiently deliver and manage power and cooling at unprecedented densities.
Implications for AI Data Centers and Future Outlook
The integration of WiseWare digital control with Navitas GaN and SiC devices represents a critical step forward in solving the power delivery bottleneck in AI data centers. By enabling real-time, dynamic adjustments to power-conversion behavior, this collaboration offers a scalable and efficient approach to managing the extreme power transients inherent in AI training and inference workloads.
Looking ahead, as the industry continues to push toward multi-megawatt AI cooling systems and increasingly dense GPU clusters, the reliance on specialized power semiconductors will only intensify. The transition from traditional electrical infrastructure to semiconductor-driven power delivery will dictate the pace of AI hardware scaling, making innovations in GaN, SiC, and digital control foundational to the future of artificial intelligence.