Can Gallium Nitride Replace Silicon Carbide?

Short answer: no, GaN will not replace SiC - but it is rapidly moving in alongside it. Thanks to its high switching frequency and low losses, gallium nitride offers more efficient and cheaper solutions than silicon carbide in the low-to-mid power range; in high-voltage traction inverters, SiC's territory remains valid. The two materials are not rivals - they are specialists of different voltage and frequency regions.

Why Does GaN Switch Faster and With Lower Losses?

The answer lies in the material's fundamental physical properties. GaN is a wide-bandgap semiconductor with a bandgap of approximately 3.4 eV ; in silicon this value is 1.1-1.12 eV, and in SiC it is 3.2 eV. A wide bandgap means the semiconductor can withstand a much higher electric field within its layer: the critical breakdown field in GaN is approximately 3.3 MV/cm, while in silicon it is only 0.3 MV/cm. The practical result is this - a GaN device can block the same voltage with a much thinner layer, which brings lower on-resistance and lower capacitance.

The second trump card is electron mobility. In a structure known as HEMT, GaN creates a highly mobile channel near the surface called a two-dimensional electron gas; mobility reaches the range of 1500-2000 cm2/Vs . In SiC the same value is 700-1000, and in silicon it is 1450. High mobility means carrying the same current with less resistance, which improves both conduction losses and the switching limit.

There is also a structural reason for the reduction in losses: GaN FETs have no separate body diode. As a result, reverse recovery loss is nearly zero - whereas the reverse recovery charge of a typical SiC FET is above 85 nC. Capacitances are also low: while the gate charge in GaN stays below 1 nC-ohm, in silicon it is 4 nC-ohm; the output charge is 25 nC-ohm versus 5 nC-ohm. New GaN generations with integrated gate drivers can be switched with edge rates of up to 150 V/ns; this means approximately 82% lower losses compared to SiC and 63% lower compared to discrete GaN FETs. A measurable summary: switching energy is less than half that of SiC.

High frequency has a two-way benefit: as switches speed up, passive components such as transformers, inductors, and output capacitors shrink. Smaller magnetics, a smaller heatsink, and a more compact converter - in power electronics, volume is usually determined by the passives, not the semiconductor itself.

GaN vs SiC Comparison: Which Material Is Better Where?

Reducing the two technologies to a single criterion would be misleading; the difference between them becomes clear across four axes.

In thermal management, the table favors SiC: a material that conducts heat more than twice as well allows you to make do with a more modest heatsink at the same loss. The counterargument on the GaN side is that, thanks to its cleaner switching, the device generates less heat under comparable load. In other words, GaN handles heat worse but creates less heat; in practical design, the two approaches can end up with similar cooling volumes.

The cost story is knotted at the substrate. GaN is grown on standard silicon substrates and directly benefits from the existing 300 mm production infrastructure. SiC, on the other hand, requires a much more expensive raw material and special processes that demand temperatures above 2500 °C; in return, it delivers more chips per wafer and better thermal endurance. On a normalized scale, GaN sits at 1.3 and SiC at 2.4 - and GaN's advantage tends to grow as scale increases. The system-level calculation points in a similar direction: smaller magnetics, less cooling, and fewer components create savings beyond the device price.

Placing the three technologies side by side with the same topology in a three-phase grid converter summarizes the situation: with IGBT, a power density of 73 W/in³ and 98.3% efficiency is achieved at 20 kHz; with SiC, 170 W/in³ and 98.9% at 100 kHz; and with GaN operating at 900 V with a multilevel topology, 211 W/in³ and 99.2% at 140 kHz. The numbers also show that high frequency is not just an efficiency metric but a power density metric.

Where Is GaN Used in Electric Vehicles Today?

On the EV side, GaN's natural territory is the onboard charger (OBC), which converts grid AC power into the vehicle's battery, along with DC-DC converters. Both operate in the medium power range, and as switching frequency increases, volume gains translate directly into trunk space. In these applications below 10 kW, the two wide bandgap materials occasionally overlap - for example, in single-phase PFC stages, GaN with the totem-pole topology offers the lowest loss and highest efficiency.

The commercial turning point came in February 2025: Changan Automotive launched the first commercially available GaN-based onboard charging unit using Navitas technology. The unit achieves 96% efficiency in charging and discharging, and 6 kW/liter reports power density. Today, GaN-based OBCs can reach an efficiency range of 96-98% and can shrink system volume by 30-60%. Progress on the reliability front is also concrete: 2026-dated research shows that p-GaN gate structures with a threshold exceeding 4 V are being developed for automotive requirements; Infineon is publishing GaN device selection guides covering bidirectional switches for EV OBCs and traction inverters, and an EU-funded project is working on compact OBCs for the same goal.

Market figures confirm this momentum: the global power GaN market is expected to rise from 615 million dollars in 2025 to over 920 million dollars; EV charging systems are the main driver of this growth. The power electronics market covering both SiC and GaN is projected to exceed 4 billion dollars by 2026.

Where Does the Limit of GaN in 800V Architecture Begin?

First, let's be clear: the 800 V architecture is not a single battery voltage, but a system approach spanning a range of approximately 600-900 V . The logic is simple - when the voltage doubles for the same power, the current drops by half, which reduces cable and connector cross-sections. But this is exactly where GaN's voltage class comes into play: GaN FETs typically remain in the 600 V class, which restricts their direct two-level use in an 800 V limited data bus. High-voltage traction inverters therefore remain SiC's domain; SiC devices go up to 1200 V and beyond, covering automotive and locomotive traction, large solar farms, and grid converters.

The limit is not a hard wall, but a matter of design choice. Multilevel topologies divide the voltage seen by each switch, allowing GaN to be used even in high-voltage applications such as a 900 V DC data bus; the 99.2% efficiency in a three-phase grid converter was achieved with exactly this approach. So the future of GaN in 800 V architecture depends on whether device voltage or topological elegance will win - in the short term, the two approaches will coexist.

Is GaN Reliable in EV Applications?

The question is fair, and the answer depends on the application. An academic evaluation emphasizes that reliable GaN use in EVs depends on how converter stress accumulates in real driving cycles and how the vehicle handles this stress; in other words, the automotive qualification process has not yet fully matured. On the other hand, comprehensive reliability tests do exist, and it should not be forgotten that GaN operates differently from silicon or SiC: there are multiple device classes, such as normally conducting (D-mode) structures on the gate, and each carries its own failure mechanism.

The balanced reading is this: GaN's structural advantages - the absence of a body diode, low capacitances, integrated drivers - theoretically mean less stress and less heat. The risk is that the long-term effects of voltage surges in field conditions, temperature cycling, and driving profiles have not yet been verified with field data as extensive as SiC's. In more controlled environments like OBC and DC-DC, the picture is quickly becoming clear; in aggressive environments like the traction inverter, the qualification process still needs to be completed.

Which Direction Is Power Electronics Heading in the Coming Years?

The current agenda is grouped under three headings: AlGaN/GaN HEMT structures, vertical GaN devices, and cascode configurations. Vertical GaN is particularly important, because it is the strongest candidate that could push the voltage ceiling of existing lateral structures upward - and thus bring GaN directly into the 800 V universe. The main application drivers are EVs and 5G infrastructure.

I can summarize my expectation as follows: in the coming years, power electronics will not be the story of a single winner, but of specialists divided into regions. Silicon will keep its low-cost, low-frequency corner; SiC will remain dominant in traction inverters requiring high voltage and high temperature, and on the grid side; while GaN will rapidly gain ground in high-frequency, compact, and cost-sensitive OBC, DC-DC, and low-to-mid power inverter stages. The overlap between them in the 10 kW band requires that the engineer's choice be made not with the question "which semiconductor is better" but "which material serves which voltage, frequency, and volume target." Every new generation of battery chemistry will demand smarter converters from power electronics - and the current data shows that GaN will take its place alongside SiC, not instead of it.

Sources

• GaN Switching Frequency
• What Makes Gallium Nitride Important for Power Electronics?
• Performance and benefits of GaN versus SiC
• How Gallium Nitride Enables Smaller, More Efficient Power Supplies
• Changan Automobile Launches First Commercial GaN-Based OBC Using Navitas Technology
• GaN topology and device selection for EV OBC and traction inverter
• Revolutionising electric mobility with compact, efficient on-board chargers
• GaN Electric Vehicle Systems-A Comparative Review
• Gallium Nitride Power Devices for High Reliability Automotive
• What is an 800 V vehicle architecture?
• 400V vs 800V Charging: What You Need to Know
• SiC & GaN Transforming Modern Power Electronics
• GaN Power Semiconductor Landscape 2026
• Rise and Rise of SiC and GaN in Power Electronic Industry
• A Comparison of Silicon, Silicon Carbide, and Gallium Nitride for Power Semiconductor Devices