Next-Generation Power Device Technology Outlook — Design and Procurement Strategies Beyond SiC and GaN

As SiC and GaN establish themselves as the mainstream in industrial and automotive power semiconductors, materials science is already advancing research and development on next-generation materials. Three candidate families are drawing significant attention: Ga₂O₃ (gallium oxide), GaN on GaN (native-substrate GaN), and diamond semiconductors, each targeting volume production in the 2030s. For designers and procurement professionals, anticipating the technology transition timeline is the key to competitive advantage.

Current Generation: Characteristics and Market Trends of SiC and GaN

GaN outperforms SiC on the high-frequency Baliga figure of merit and excels at low-loss operation in high-frequency, high-voltage applications. Multiple market research firms project the GaN power semiconductor market to grow at a CAGR in the 20–50% range, driven by AI server power supplies, 800 V EV architectures, and 5G communication infrastructure. NVIDIA is advancing the shift to 800 V power architectures for AI servers, positioning GaN as a key enabling technology for high-efficiency power conversion.

SiC's high breakdown field strength and low on-resistance (Ron) have brought it into volume production for EV traction inverters. As the transition to 8-inch wafers progresses, device costs are estimated to fall 20–35% over the long term.

GaN on GaN — Advantages and Challenges of Native Substrates

GaN on Si (GaN on silicon substrate) keeps volume production costs low, but the high dislocation density caused by lattice mismatch with the substrate affects reliability. GaN on GaN (GaN on GaN substrate) substantially reduces dislocation density, making it easier to overcome reliability issues such as charge trapping and gate instability. The short-circuit withstand time (SCWT) of GaN on Si — well below 1 μs and far shorter than silicon IGBTs — is also expected to be less constrained on native substrates.

GaN single-crystal substrate manufacturing costs remain high, and progress toward larger diameters has been slow, with 2–4 inches remaining the norm today. Achieving high-quality, large-diameter, low-cost GaN substrates is the primary technical barrier to volume adoption; Sumitomo Electric and Mitsubishi Chemical are among the companies developing volume production technologies. Multiple device makers developing AEC-Q101-qualified GaN on Si devices are targeting volume production around 2027–2028, a timeline that has become a broad industry consensus, with migration to native substrates as a subsequent option.

Ga₂O₃ (Gallium Oxide) — A Leading Candidate for Ultra-High Voltage

β-Ga₂O₃ has a bandgap of approximately 4.8 eV, substantially exceeding SiC (3.3 eV) and GaN (3.4 eV), and its breakdown field strength reaches approximately three to four times that of SiC. In theory, it offers outstanding performance for ultra-high-voltage applications above 10 kV — power grids, railways, and large industrial equipment.

Two major challenges currently impede volume production: difficulty in p-type doping and low thermal conductivity (approximately one-quarter that of SiC). Substrate supply businesses, including Novel Crystal Technology (Japan), are beginning to emerge, and partial commercialization for industrial applications in the early 2030s is considered a realistic outlook.

Diamond Semiconductors — Distance to Practical Use

Diamond has the highest bandgap (5.5 eV) of all semiconductor materials, the highest thermal conductivity (approximately five times that of SiC), and high carrier mobility, enabling operation in extreme environments — ultra-high temperatures, space, and defense. AIST and Sumitomo Electric are advancing prototype demonstrations, but the difficulty and cost of synthesizing large-diameter single crystals means practical volume production devices are not expected before the late 2030s to 2040s.

Technology Maturity and Volume Production Outlook

Implications for Designers and Procurement Professionals