Comparing Investment Strategies for SiC, GaN, and IGBT by Major Semiconductor Manufacturers

The way to bet on power semiconductors is starting to diverge in 2025

A shift is happening in the main technology for EV inverters, pressure to improve efficiency in industrial equipment, and a surge in power density for data centers – multiple waves of demand are hitting simultaneously. Major semiconductor manufacturers are now being forced to choose where to invest capital, where to defend, and where to abandon among three technologies: SiC, GaN, and IGBT.

Before answering this question, let's look at one number. Multiple research institutions indicate that the SiC power semiconductor market will grow at a compound annual rate of over 30% towards 2030, with Infineon and STMicroelectronics announcing manufacturing investments in the hundreds of billions of yen. On the other hand, IGBTs still support a large portion of the world's power converters, and GaN is solidifying its niche in high-frequency, low-voltage domains. What lies before us is a complex investment landscape where one cannot definitively say "SiC only" or "do all three materials."

This article will delve into how major players are actually allocating capital and development resources, after covering the fundamentals of the technology. We hope this will serve as a blueprint for design selection, negotiation material for procurement, and a framework for business decisions.

The three materials should be viewed in terms of "coexistence" rather than "competition."

To clarify the discussion, let's first confirm the positioning of the three materials. IGBTs (Insulated Gate Bipolar Transistors) are silicon devices that combine the characteristics of MOSFETs and bipolar transistors, playing a key role in power conversion from hundreds of volts to several kilovolts. SiC (silicon carbide) is a wide-bandgap semiconductor with an electric breakdown field approximately 10 times higher than silicon, capable of significantly reducing losses under high voltage, high temperature, and high frequency conditions. GaN (gallium nitride) is also a wide-bandbandgap material, but its high electron mobility makes it particularly adept at high-frequency switching, and it is gaining prominence in onboard chargers (OBCs) and power modules.

When we organize the three materials by "voltage range × frequency range," the coexistence becomes quite clear.

Despite the clear coexistence, competition does exist. In the 600 to 900V range, SiC and IGBTs compete for the same applications. Factory solar power conditioners and industrial motor drives are typical examples where the choice between continued IGBT selection or replacement by SiC hinges on cost. As we will see in the next section, this "replacement competition" is shaping the investment strategies of each company.

Focus on SiC or broaden the portfolio? — Divergence in major players' strategies

An overview of major manufacturers' efforts in SiC, GaN, and IGBT reveals two main directions: "SiC-focused" and "comprehensive three-material approach."

Infineon is the largest player with both aspects. While offering SiC devices and modules under the CoolSiC brand, they continue to strengthen IGBT technology. They have also entered the GaN market with CoolGaN products, and their large business scale provides the financial strength to cover all three materials simultaneously. In manufacturing, they are investing in the transition to 300mm wafers in Malaysia and Dresden, demonstrating a strategy to differentiate from competitors by reducing costs through economies of scale.

STMicroelectronics is known for its early focus on SiC. Starting with a large supply agreement for SiC inverter modules to Tesla, they have solidified their position in the EV supply chain. In addition to their dedicated SiC factory in Catania, Italy, they have also established a production system in China through a joint venture with Sanan Optoelectronics. ST's case illustrates how long-term supply contracts with EV manufacturers can drive investment decisions. The SiC module supply agreement between ST and Ampere mentioned at the beginning of the article (starting in 2026) is also part of this trajectory.

onsemi offers a full lineup of SiC MOSFETs, diodes, and modules from 650V to 1700V under the EliteSiC brand. They are accelerating vertical integration by combining in-house wafer manufacturing through the acquisition of Globalfoundries with substrate procurement agreements from GT Advanced Technologies.

Mitsubishi Electric, while having a long track record in IGBTs, continues to invest in the transition to SiC. Their efforts to improve short-circuit withstand time using a p-type protective layer in trench SiC MOSFETs are technically noteworthy and offer one answer to the short-circuit withstand time issue discussed later.

ROHM is a representative Japanese manufacturer with a strong focus on SiC. Their fourth-generation SiC MOSFETs aim to achieve both low on-resistance and high short-circuit withstand time through a unique structure, and their integrated production system from wafer to epitaxy to devices is also valued from a supply chain risk perspective.

The "hidden hurdle" in SiC selection: short-circuit withstand time in practical applications

When considering the adoption of SiC devices, judging solely by on-resistance (RonA) and breakdown voltage in the specifications can lead to problems later. One reason for this is the short-circuit withstand time (SCWT).

Load short circuits can occur due to various causes, including inverter control malfunctions, wiring errors, or hardware failures. During this time, lasting a few microseconds until the protection circuit operates, the device absorbs energy far exceeding its rating. Because SiC has small dies and high current density, it heats up faster than silicon devices, making the response requirements for protection circuits more stringent.

Microchip's 700V/1200V products specify a SCWT of typ. 3μs under specific conditions. This figure directly impacts protection circuit design, requiring the entire circuit to be designed in conjunction with the detection time of the gate driver IC's DESAT (desaturation) function and the blanking time settings.

DESAT is a function that monitors the drain-source voltage (VDS) in the on-state and detects the voltage rise that occurs during overcurrent to turn off the power transistor. The reliability of the protection operation hinges on three parameters: the DESAT trigger threshold (VDESAT), the DESAT current (IDESAT), and the blanking time.

Short-circuit withstand time also depends on operating conditions such as drain applied voltage, gate voltage, and junction temperature. At higher temperatures, RDSon increases, suppressing saturation current and thus tending to improve short-circuit withstand capability. Conversely, margin confirmation under low-temperature, high-voltage conditions becomes an important confirmation item in circuit design.

The trade-off between short-circuit withstand time and on-resistance in SiC is a well-known fact in the industry, but how each company is technically trying to resolve this trade-off is a valuable criterion for device selection. Mitsubishi Electric improves the short-circuit withstand time of its trench-type devices with a p-type protective layer, while Rohm aims to achieve both low RonA and high short-circuit withstand time with its fourth-generation structure. Determining which approach is suitable for your company's application during the design phase can lead to reduced development risks later on.

Are IGBTs "obsolete"? — How to interpret the continuous evolution of silicon technology

"As SiC replacement progresses, won't IGBTs just shrink?" — Some hold this view. However, the reality is somewhat more complex.

Toshiba Device & Storage's triple-gate IGBT, announced in 2021, achieved a loss reduction of up to 40.5%. This is a prime example cited repeatedly in the industry, demonstrating that silicon devices still have room for performance improvement. The cost advantage of IGBTs remains strong, particularly in large-capacity, high-voltage industrial inverters, and a complete transition to SiC is unlikely in the short term for applications like machine tools, railways, and wind power generation.

This graph illustrates the simple fact that the higher the voltage range, the stronger the presence of IGBTs. For power grids above 6.5kV, the product range that can be replaced by SiC is currently limited, and the market for IGBT manufacturers will not disappear anytime soon.

However, caution is needed in the 1200 to 1700V range. This is a major battleground for SiC, and the replacement competition in industrial inverters and large solar power conditioners is currently underway. To maintain IGBTs in this range, either performance improvements like the triple-gate structure or thorough cost reductions are required. Several manufacturers are pursuing both approaches simultaneously.

Why is GaN not just "the third option"?

GaN is often positioned as a "complementary material to SiC," but looking at its market growth trajectory, it has independent growth drivers.

EV onboard chargers (OBCs) are one of GaN's main markets. The demand for smaller and more efficient chargers is increasing year by year, and GaN's characteristics, enabling switching frequencies of hundreds of kHz or more, directly translate to a competitive advantage. Furthermore, the transition to 48V power supply architectures in data centers is also boosting demand for GaN power ICs.

Renesas Electronics' acquisition of Transphorm's GaN business can be seen as a response to this trend. By incorporating GaN into their portfolio as a wide-bandgap material following SiC, they aim to provide customers with the ability to select devices based on voltage range and application.

GaN on Si manufacturing can utilize existing silicon production lines in part, potentially making it more cost-effective than SiC in terms of wafer costs. However, reliability evaluations specific to GaN devices—particularly degradation of dynamic on-resistance and warpage of GaN on Si substrates—remain elements that need to be considered during the design phase.

This graph helps to organize the differences in characteristics: SiC and GaN have equivalent potential in dielectric breakdown field, while SiC excels in thermal conductivity and GaN surpasses it in electron mobility. High-temperature, high-power density applications (EV inverters, industrial equipment) benefit from SiC's thermal conductivity, while high-frequency, medium-voltage applications (OBCs, power supplies) leverage GaN's mobility—this difference forms the basis for coexistence.

How to read the investment map — Examining technology, manufacturing, and supply chain layers

Let's organize the strategies of each company by three layers: "technology," "manufacturing," and "supply chain."