GaN Talk Blog

The full article was originally published in EPDT

The history of power electronics has been a predictable cadence of material limits, architectural shifts, and a new semiconductor platform resetting expectations. Today that transition is happening again, this time with gallium nitride (GaN).

AI’s rapid rise and the widespread rollout of data centers needed to sustain its growth are fundamentally changing the way data centers are designed—not only in terms of compute density and cooling requirements, but also in how electrical power is delivered. Racks that once drew tens of kilowatts are now approaching hundreds of kilowatts, with megawatt-scale AI compute clusters no longer a distant target. In this context, traditional power architectures based on 415/480 VAC distribution and low-voltage DC buses are becoming increasingly inefficient, bulky, and expensive to scale.

A partnership with Renesas, which offers high-voltage GaN, suggests an opportunity for EPC to broaden its low- and medium-voltage portfolio.

Gallium Nitride (GaN)—a wideband gap semiconductor alongside silicon carbide (SiC)—has emerged as a superior alternative to silicon in next-generation power electronics. Yole—a renowned market research firm—forecasts $ 3 billion in 2030 for the GaN power market, with a remarkable compound growth of 42% between 2024 and 2030, as a result of OEM uptake, consumer maturity, and NVIDIA-supported AI datacenter business expansion.

Power Electronic News
Read article

In an interview with Judith Cheng, Editor at Large at MakerPROTW, Alex Lidow, co-founder and CEO of Efficient Power Conversion (EPC), described the company’s seventh-generation gallium nitride (GaN) technology, now in mass production, and its implications for low-voltage applications that have traditionally been dominated by silicon MOSFETs. With devices such as the 40 V EPC2366 already in volume production, EPC is positioning GaN as a mainstream option for 40 V and below - a market that exceeds the size of the 100 V segment where GaN first gained traction.

Michael A. de Rooij and Alejandro .P. Pozo

This article orginally appeared on EEPower.

Artificial intelligence is now in vogue - and its influence will only continue to grow. Today, the real challenge lies in delivering the energy demands at optimal efficiency levels. NVIDIA recently released a white paper¹ showing that an 800‑volt DC architecture combined with energy storage solutions represents an alternative approach to meeting the needs of modern and future AI server infrastructures. These systems will require power delivery in megawatts, not just kilowatts. The enabling technology is GaN. In collaboration with NVIDIA, EPC has developed a cost-effective, low-profile 800 VDC‑to‑12.5 VDC converter² capable of delivering up to 6 kW. The design leverages EPC’s latest 150 V and 40 V GaN device^3,4, fitting within a footprint of 5,000 mm² and a total thickness of only 8 mm. The design exemplifies how GaN FETs can achieve a combination of high efficiency and high power-density while keeping costs low, aiming to achieve 97% efficiency at full load.

At the Bodo Power Systems Wide Band Gap Forum in Monaco, EPC founder and CEO Alex Lidow set the tone for the GaN discussion, highlighting the significant advantage of gallium nitride (GaN) by drawing on fifty years of experience in power semiconductors. Speaking alongside experts from Texas Instruments, Navitas, Infineon, Toshiba, Volkswagen, and Mitsubishi, he positioned GaN as the superior choice for low-voltage, high-frequency systems—from AI data centers and humanoid robots to autonomous vehicles and LiDAR. While SiC remains the go-to for high voltages, Lidow highlighted GaN as the cost-effective technology already reshaping point-of-load power—and much more.

48 V is being adopted in many applications, including AI systems, data centers, and mild hybrid electric vehicles. However, the conventional 12 V ecosystem is still dominant, so a high power density 12 V to 48 V boost converter is required. The fast-switching speed and low R_DS(on) of eGaN FETs can help address this challenge. In this post, the design of a 12 V to 48 V, 500 W DC-DC power module using eGaN^® FETs directly driven by eGaN FET compatible ISL81807 controller IC from Renesas in the simple and low-cost synchronous boost topology is evaluated.

The latest generation of 100 V GaN devices increase the efficiency, shrink the size, and reduce system cost for 48 V power conversion. The EPC2045, shown in figure 1, is rated at 100 V with 7 mΩ on- resistance that can carry a continuous current of 16 A. The EPC2045 is nearly one-tenth the footprint of a comparable Si MOSFET and has lower parasitic capacitances and can switch much faster than equivalent silicon devices, yielding lower switching loss even at higher switching frequency.

The EPC2053, shown in figure 2, is rated at 100 V with 4 mΩ on-resistance that can carry a continuous current of 32 A. The EPC2053 has lower parasitic capacitances and on-resistance than its silicon counterparts, yielding faster switching speed and lower power losses even at higher switching frequencies. These characteristics enable increasing the output power while shrinking the volume of the converter.

Motivation

The rapid expansion of the computing and telecommunication market is demanding an ever more compact, efficient and high power density solution for intermediate bus converters. The LLC resonant converter is a remarkable candidate to provide a high power density and high-efficiency solution. eGaN^® FETs with their ultra-low on-resistance and parasitic capacitances, benefit LLC resonant converters by significant loss reduction that is challenging when using Si MOSFETs. A 48 V to 12 V, 900 W, 1 MHz LLC DC to DC transformer (DCX) converter employing eGaN FETs such as EPC2053 and EPC2024 is demonstrated, yielding a peak efficiency of 98.4% and a power density exceeding 1500 W/in³.

Gallium nitride (GaN) is a better semiconductor than silicon. There are many crystals that are better than silicon, but the problem has always been that they are far too expensive to be used in every application where silicon is used. But, GaN can be grown as an inexpensive thin layer on top of a standard silicon wafer enabling devices that are faster, smaller, more efficient, and less costly than their aging silicon counterparts.