-
Navitas Semiconductor announced its participation in NVIDIA’s Partner Ceremony held on May 29, 2026, at the Taipei Nangang Exhibition Center. The event brought together ecosystem partners supporting the NVIDIA AI Factory MGX™ platform, highlighting industry collaboration aimed at advancing next-generation AI data centers based on emerging 800 VDC rack architectures.
As part of the NVIDIA AI Factory MGX™ Ecosystem Showcase at COMPUTEX 2026 in Taipei, taking place from June 2–5, Navitas is demonstrating its 800 V-to-6 V DC-DC power delivery board (PDB). Powered by the company’s GaNFast™ technology, the PDB eliminates the need for a traditional 48 V intermediate bus converter (IBC) stage within compute server trays, enabling greater system efficiency, reliability, and board-space utilization.
The power delivery board incorporates sixteen 650 V, 11 mΩ GaNFast FETs in the company’s latest DFN8×8 dual-cooled package. The design targets 97.5% peak efficiency while operating at a switching frequency of 1 MHz and achieving a power density of 2,100 W/in³. The ultra-low-profile design is approximately 20% thinner than a mobile phone, allowing close integration with GPU boards to enhance transient response and improve power distribution efficiency.
“As AI workloads continue to scale and drive unprecedented demand for compute, power delivery has become one of the most critical challenges in enabling next-generation gigawatt AI factories,” said Chris Allexandre, President and CEO of Navitas. “Through our collaboration with NVIDIA within the MGX™ ecosystem, Navitas is delivering GaN and SiC power technologies that enable megawatt-scale AI server racks with higher power density, a smaller system footprint, and improved thermal performance, helping accelerate the transition to more efficient and scalable AI infrastructure.”
Navitas also highlighted its portfolio of wide-bandgap power technologies designed to support next-generation AI factory infrastructure. The company’s GeneSiC™ silicon carbide (SiC) solutions address power delivery requirements from the electrical grid to AI compute racks, supporting applications such as solid-state transformers (SSTs) with 2300 V and 3300 V SiC power modules, as well as high-power three-phase power supply units based on its fifth-generation 1200 V SiC MOSFET technology.
According to the company, these SiC technologies contribute to improved efficiency, increased power density, and enhanced system reliability in large-scale AI data center deployments.
Navitas’ GaNFast technology is designed to provide the high-frequency, high-efficiency DC-DC power conversion required to support increasing power demands from AI accelerators and GPUs. By leveraging the switching performance of gallium nitride devices, the company’s solutions enable MHz-frequency operation, higher power density, and faster transient response, facilitating more efficient power delivery from the rack level to the processor level.
Through its portfolio of GaN and SiC technologies, Navitas continues to collaborate with NVIDIA within the MGX ecosystem to support open and modular AI infrastructure architectures and contribute to the development of next-generation AI factory platforms.
Original – Navitas Semiconductor
-
Power Integrations has introduced two new ultra-slim auxiliary power supply reference designs developed for 800 VDC AI data center applications. The designs are optimized for NVIDIA’s Kyber liquid-cooled blade-rack architecture and are intended to reduce space requirements, simplify system design, and improve reliability in high-voltage AI infrastructure.
The first design is a single-output 15 W auxiliary power supply measuring just 30 mm × 30 mm with a profile height of 7 mm. The second is an isolated six-rail 35 W design measuring 80 mm × 60 mm with an 8 mm profile. According to Power Integrations, the compact solutions can free up approximately 30% of board space on densely populated main power distribution boards (PDBs) while reducing bill-of-materials (BOM) count by an estimated 30%, contributing to simplified designs and enhanced system reliability.
Both reference designs achieve efficiency levels of at least 88% across line and load conditions.
“As the only company offering single-HEMT 1700 V GaN devices, Power Integrations can design these best-in-class, highly efficient flyback converters with a low BOM count while maintaining wide safety margins on an 800 V bus,” said Jason Yan, Senior Training Manager at Power Integrations. “The only alternative solutions are discrete, costly silicon carbide (SiC) devices which require 30% more components and space to operate.”
The newly released design examples describe 35 W and 15 W flyback auxiliary power supplies intended for high-voltage AI data center environments. These compact power supply units provide power for internal system components such as microcontrollers, gate drivers, and operational amplifiers that perform essential control, monitoring, and housekeeping functions required for system safety, reliability, and efficiency.
Both designs are based on Power Integrations’ InnoMux™-2 ICs incorporating 1700 V-rated PowiGaN™ gallium nitride technology. The InnoMux-2 devices support nominal input voltages up to 1000 VDC in a flyback topology and can deliver up to 90% flat efficiency in discontinuous conduction mode (DCM) while maximizing power delivery performance.
Power Integrations has made both reference designs available for download:
• DER-1110 – a 35 W multi-output flyback auxiliary power supply based on the IMX2353F, designed for high-voltage AI data center applications.
• DER-1114 – a 15 W single-output flyback auxiliary power supply based on the IMX2353F, also targeted at high-voltage AI data center applications.
The new designs leverage Power Integrations’ 1700 V PowiGaN technology to deliver compact, low-profile auxiliary power solutions for emerging 800 VDC AI data center architectures.
Original – Power Integrations
-
Efficient Power Conversion (EPC) has released additional information on its EPC91123 evaluation board, a 6 kW isolated converter designed to convert 800 VDC to 12.5 VDC for next-generation AI data center power architectures.
As a contributor to the NVIDIA MGX™ AI Factory ecosystem, EPC is advancing gallium nitride (GaN)-based power conversion technologies intended to support emerging 800 VDC server architectures. The company’s solutions are designed to enable higher power density, improved efficiency, and scalable rack-level power delivery for future AI infrastructure.
The EPC91123 enables direct conversion to 12.5 V, eliminating the need for a traditional 48 V intermediate bus stage. According to EPC, this approach improves overall system efficiency by approximately 1–2%.
The converter is based on an input-series, output-parallel (ISOP) LLC topology and incorporates EPC’s Gen7 eGaN® devices, including the EPC2366, a 40 V device with 0.8 mΩ on-resistance in a compact 3.3 mm × 2.6 mm package, and the EPC2305, a 150 V device with 2.2 mΩ on-resistance.
The EPC91123 achieves a peak efficiency of 98.2% and a full-load efficiency of 97%. Designed for space-constrained AI server environments, the evaluation board delivers high power density within a compact 104 mm × 47 mm × 8 mm form factor.
“AI data centers consume tremendous amounts of power, making it critical to reduce power conversion stages to improve efficiency and power density, especially in 800 V architectures,” said Alejandro Pozo, Director of DC-DC System Engineering. “The ISOP topology is particularly well suited for high step-down-ratio power conversion, enabling improved transformer optimization and interleaved operation. It has been selected for our next 800VDC to 6VDC platform.”
The EPC91123 forms part of EPC’s broader strategy for 800 VDC AI power delivery systems. The company is developing conversion platforms spanning 800 VDC to 48 VDC, 12 VDC, and 6 VDC to address a range of system-level requirements in AI infrastructure.
By utilizing advanced GaN-based ISOP topologies and its latest Gen7 eGaN technology, EPC aims to provide scalable, high-density power conversion solutions that improve efficiency, reduce power distribution losses, and support next-generation accelerated computing platforms.
“NVIDIA MGX provides a modular foundation for scalable accelerated computing. A key element of NVIDIA’s strategy for AI infrastructure is the use of 800 VDC power architecture, helping address the growing demands for efficiency and power density as AI compute scales. We are pleased to contribute to the NVIDIA MGX AI Factory ecosystem with advanced multi-level GaN-based power conversion solutions designed to support emerging 800 VDC server architectures and next-generation AI infrastructure,” said Alex Lidow, CEO and Co-founder of EPC.
Additional information on the EPC91123 evaluation board, EPC’s Gen7 GaN technology, and the company’s portfolio of 800 VDC power conversion solutions for AI infrastructure is available from EPC.
Original – Efficient Power Conversion
-
onsemi is expanding its participation in the NVIDIA MGX™ ecosystem, strengthening its position as a key supplier of power semiconductor technologies for next-generation AI infrastructure.
As AI workloads drive rack power requirements toward and beyond 1 MW, efficient power delivery has become one of the most critical challenges facing hyperscale and enterprise data centers. NVIDIA’s MGX modular architecture is designed to accelerate deployment of AI systems, and onsemi is providing power solutions across multiple stages of the power conversion chain within this ecosystem.
The company’s portfolio spans silicon, silicon carbide (SiC), and gallium nitride (GaN) technologies, supporting applications ranging from power supply units (PSUs) and battery backup units (BBUs) to emerging 800 VDC power distribution architectures. Existing MGX platforms already utilize onsemi technologies, with opportunities expanding across power MOSFETs, multiphase power solutions, SiC JFETs, and GaN devices.
A major focus is the industry’s transition toward high-voltage 800 VDC architectures, which are increasingly viewed as essential for scaling AI infrastructure efficiently. These architectures enable higher power density, reduced distribution losses, lower infrastructure complexity, and support for future rack-level power requirements exceeding 1 MW.
Original – onsemi
-
Infineon Technologies AG has joined NVIDIA’s MGX AI Factory ecosystem, strengthening its role in next-generation AI data center power infrastructure and supporting the industry transition toward 800 VDC power distribution architectures.
Through the collaboration, Infineon will provide power management technologies for NVIDIA’s MGX™ modular server platform and emerging 800 VDC data center architecture. The partnership focuses on improving power efficiency, power density, and scalability as AI workloads continue to drive unprecedented energy demands across hyperscale data centers.
Infineon’s contribution spans the entire power conversion chain from grid to processor core, leveraging its portfolio of silicon (Si), silicon carbide (SiC), and gallium nitride (GaN) technologies. The company’s GaN devices enable ultra-compact, high-frequency bus converters operating near 1 MHz, while its proprietary SiC JFET technology supports critical protection and hot-swap functions for native 800 V server platforms.
The company’s power solutions support multiple conversion stages, including 800 V to 50 V, 12 V, and 6 V architectures, helping reduce conversion losses and move power delivery closer to AI processors. This approach improves overall system efficiency while enabling higher rack power densities required by future AI infrastructure.
Original – Infineon Technologies
-
Innoscience outlined its roadmap for enabling next-generation AI data center power delivery through an end-to-end “All-GaN” architecture aligned with the evolving NVIDIA MGX ecosystem and emerging 800 VDC power distribution standards.
As AI factories scale toward megawatt-level rack power, efficient power conversion has become a critical bottleneck. Innoscience positions gallium nitride (GaN) technology as a key enabler of future AI infrastructure, citing its ability to deliver higher switching frequencies, lower losses, improved thermal performance, and greater power density than conventional silicon solutions.
The company’s roadmap spans the entire AI power delivery chain, from 800 VDC rack distribution down to GPU core voltages. At the front-end conversion stage, Innoscience demonstrated a 12 kW 800 V-to-48 V all-GaN LLC converter utilizing 650 V and 100 V GaN devices, achieving approximately 99% peak efficiency and 98.2% full-load efficiency while operating at 1 MHz. The company has also introduced 150 V GaN devices that reduce secondary-side synchronous rectifier requirements by 50%, simplifying system design and improving power density.
Beyond the 48 V architecture, Innoscience is extending its portfolio to support emerging 800 V-to-12 V and 800 V-to-6 V conversion topologies that are gaining traction in next-generation AI server designs. These architectures aim to reduce conversion stages, lower distribution losses, and move power conversion closer to GPUs.
For intermediate bus conversion, the company highlighted its 100 V GaN portfolio for 48 V-to-12 V multiphase buck converters, targeting higher efficiency and power density in AI servers where even marginal efficiency gains can significantly reduce cooling and operating costs at data center scale.
At the point-of-load level, Innoscience is developing 15 V DrGaN technology for vertical power delivery (VPD) architectures. Operating at switching frequencies between 3 MHz and 5 MHz, these solutions are designed to support future GPU power requirements by reducing passive component size, improving transient response, and enabling power stages to be located closer to AI accelerators.
Original – Innoscience Technology
-
Analog Devices outlined its role in enabling next-generation AI factory power infrastructure through collaboration within the NVIDIA MGX™ ecosystem, focusing on the transition from traditional 48 V architectures to 800 VDC power distribution.
As AI workloads continue to drive higher rack-level power densities, conventional power delivery systems are reaching their practical limits. ADI emphasizes that future AI factories will increasingly adopt 800 VDC architectures to reduce current levels, lower distribution losses, improve power density, and support megawatt-scale rack deployments required by advanced AI clusters.
The company highlighted the growing trend toward modular power architectures enabled by NVIDIA MGX, where power systems are increasingly disaggregated from compute racks. This approach allows greater design flexibility, faster technology adoption, and more efficient utilization of rack space for compute resources.
A critical element of this transition is high-voltage hot-swap capability, which enables live insertion and removal of server trays without disrupting operations. At 800 VDC, this requires sophisticated protection and control technologies capable of managing high-energy inrush currents, detecting faults rapidly, and providing real-time telemetry.
ADI is leveraging its expertise in power protection, monitoring, and control to develop 800 VDC hot-swap and DC-DC conversion solutions tailored for AI infrastructure. These technologies are designed to provide high-density integration, fast fault protection, and advanced telemetry functions that support system optimization and predictive maintenance.
Original – Analog Devices
-
Innoscience has announced its collaboration with NVIDIA to support the implementation of the 800 VDC rack power architecture, a major breakthrough in power distribution for AI data centers. Similar to the transition from 400 V to 800 V systems in electric vehicles, this new architecture dramatically increases efficiency, power density, and sustainability by reducing current flow sixteenfold compared to traditional 48 V systems. The result is significantly lower resistive losses, reduced copper consumption, and lower overall CO₂ emissions.
Current 48 V data center systems are approaching physical and thermal limits, with high copper demand and nearly half of total power lost to heat dissipation. The 800 VDC system overcomes these constraints, enabling scalability from kilowatt to megawatt levels for future AI clusters with hundreds of GPUs. However, achieving such high power density and efficiency—especially converting from 800 V down to 1 V—requires advanced semiconductor technologies such as gallium nitride (GaN).
Innoscience’s third-generation GaN devices are designed to meet the demanding requirements of this new power paradigm. Operating at switching frequencies near 1 MHz, these devices significantly reduce the size of magnetic and capacitive components while improving overall system efficiency.
Key advantages of Innoscience’s third-generation GaN include:
- Up to 80% reduction in driver losses and 50% lower switching losses compared to SiC devices, yielding a 10% decrease in total system power loss.
- Twice the power density on the 54 V output side, requiring only 16 GaN devices instead of 32 Si MOSFETs for equivalent conduction performance.
- 70% reduction in switching losses and a 40% increase in power output compared to silicon-based architectures.
- Scalability to support next-generation GPU platforms with improved dynamic response and reduced capacitor costs.
As the world’s only integrated GaN manufacturer (IDM) producing devices from 1200 V to 15 V, Innoscience offers a complete GaN-based conversion chain from 800 V to 1 V, delivering an all-GaN power solution across every stage of the data center power architecture.
Innoscience’s GaN devices are also proven for long-term reliability, passing extended high-temperature and stress tests, including 2000-hour dynamic HTOL and 175°C endurance validation. These tests confirm datacenter-grade durability with expected lifetimes exceeding 20 years.
With its third-generation GaN technology and collaboration with NVIDIA, Innoscience is helping drive the shift from kilowatt to megawatt AI racks—ushering in a new era of ultra-efficient, high-performance, and sustainable AI computing infrastructure.
Original – Innoscience Technology
-
Navitas Semiconductor has announced new progress in developing high-performance gallium nitride (GaN) and silicon carbide (SiC) power devices designed to enable NVIDIA’s 800 VDC data center architecture for next-generation AI factory computing platforms.
With AI workloads driving data center power consumption into the megawatt range, traditional 54 V power distribution systems are reaching their limits. The shift to 800 VDC architecture offers higher efficiency, reduced copper use, simplified thermal management, and compatibility with global low-voltage DC standards. The new system allows direct conversion from 13.8 kVAC utility power to 800 VDC, eliminating multiple conversion stages and improving overall system reliability.
Navitas’ GaNFast™ and GeneSiC™ technologies power every stage of the AI data center — from grid connection to GPU-level conversion. The company’s new 100 V GaN FETs deliver high power density and thermal performance in dual-sided cooled packages, optimized for DC-DC conversion on GPU boards. Manufactured using a 200 mm GaN-on-Si process in partnership with Power Chip, these devices support scalable, high-volume production.
In addition, Navitas’ 650 V GaN portfolio introduces new high-power GaN FETs and GaNSafe™ power ICs, which integrate drive, sensing, protection, and control functions for enhanced safety and robustness. GaNSafe™ features ultra-fast short-circuit protection, 2 kV ESD tolerance, programmable slew-rate control, and a simple 4-pin configuration for easy implementation.
Navitas’ GeneSiC™ SiC technology, based on its proprietary trench-assisted planar structure, provides high-speed, cool-running performance and wide voltage coverage up to 6,500 V. These devices are already being used in grid-tied inverters, large-scale energy storage systems, and U.S. Department of Energy (DoE) projects.
“As NVIDIA drives transformation in AI infrastructure, we’re proud to support this shift with GaN and SiC power solutions that enable the efficiency, scalability, and reliability required by next-generation data centers,” said Chris Allexandre, President and CEO of Navitas. “The move from legacy 54 V systems to 800 VDC is not just an evolution — it’s a complete transformation of data center power.”
Navitas’ innovations in GaN and SiC technologies reflect its expanded focus beyond mobile and consumer markets toward powering megawatt-scale AI factories, industrial platforms, and smart energy infrastructure with high-efficiency, high-density semiconductor solutions.
Original – Navitas Semiconductor
-
ROHM has published a new white paper exploring advanced semiconductor solutions that enable the next generation of AI data centers powered by the 800 VDC architecture. The paper highlights ROHM’s role as a leading innovator in wide bandgap technologies and system-level power design for large-scale, energy-efficient computing infrastructure.
Developed as part of ROHM’s ongoing collaboration with NVIDIA and other industry partners, the white paper outlines strategies for implementing 800 VDC power distribution across AI data centers—a shift expected to transform data center design by enhancing efficiency, scalability, and sustainability.
The 800 VDC architecture supports the evolution of gigawatt-scale AI factories by significantly improving power density and reducing conversion losses. ROHM’s broad portfolio of silicon (Si), silicon carbide (SiC), and gallium nitride (GaN) power devices, combined with its analog IC expertise, positions the company to deliver optimized solutions for each stage of power conversion.
The paper details how the traditional AC-DC conversion process, typically performed inside server racks, can be relocated to a centralized power rack under the 800 VDC system. This approach simplifies design, improves thermal performance, and allows higher-density configurations for GPU-heavy AI workloads.
ROHM’s EcoSiC™ and EcoGaN™ device families are featured prominently as key enablers of this transition. The EcoSiC™ series provides industry-leading low on-resistance and top-side cooling modules for high-power AI servers, while the EcoGaN™ series integrates GaN performance with ROHM’s proprietary analog IC technologies, including Nano Pulse Control™, enabling fast, precise, and stable high-frequency operation.
Through collaborations with NVIDIA, Delta Electronics, and data center operators, ROHM continues to drive innovation in wide bandgap semiconductors and analog control technologies. Its new white paper underscores the company’s commitment to building efficient, reliable, and sustainable AI data center infrastructure based on 800 VDC power delivery systems.
Original – ROHM
