The semiconductor industry is at a point where performance, efficiency, and reliability must advance in parallel. The needs of AI infrastructure, electric mobility, power conversion, and communications systems are pushing materials to their limits. Gallium Nitride (GaN) is gaining attention because it can meet those needs. The industry has reached a point where the conversation is no longer if GaN is viable, but how to deploy it reliably and at scale.
Having spent over two decades focused on epitaxial growth, I’ve watched GaN move from a niche, research-driven material into a leading contender in power electronics. The progress has been steady, not overnight. The companies and engineers leaning into GaN now are the ones positioning themselves for the next generation of systems.
Device performance begins with epitaxy
For GaN, the epitaxy defines the device performance. That’s true in compound semiconductors generally, but with GaN, it’s not just important — it’s hypercritical. The reason is simple: GaN and silicon were never meant to go together. They’re highly mismatched in both lattice structure and thermal behavior. You can’t ignore that and expect the material stack to function as intended.
Growth initiation is the first critical step. In RF applications, for instance, if the epitaxial layer doesn’t interface cleanly with the substrate, you risk parasitic effects that degrade performance. For power devices, thermal management becomes a challenge if defectivity isn’t tightly controlled. Wafer flatness, morphology, and defect density all feed into long-term reliability.
This is where a lot of the real engineering takes place — not just in achieving good results once, but in doing it repeatedly on a large scale. You need consistency, not just performance. If the process can’t scale, the technology doesn’t scale.
GaN on Silicon will be performance and cost champion
GaN on silicon will lead the way in power electronics because it can meet both performance and cost targets. From a substrate standpoint, silicon has a clear cost advantage over silicon carbide (SiC). It’s widely available, supports larger wafers, and utilizes infrastructure that’s already well-established. That’s important for companies that are designing for scale and looking for predictable economics.
From a performance perspective, GaN enables higher switching speeds, better thermal behavior, and reduced conduction losses. These translate into smaller systems, lower energy use, and less demand on thermal management — all of which matter in power-dense applications.
The combination of cost structure and electrical performance makes GaN on silicon a practical and scalable choice. It’s not theoretical. It’s already being used in systems that need efficiency and reliability at volume.
GaN supply chains need new approach
Building with GaN means dealing with new manufacturing challenges. The investment required to install the infrastructure for epitaxy alone is substantial. The skill set is specialized. And the risk of quality variation is high if you’re not experienced with the process.
This is one reason GaN requires a rethink of the traditional supply chain. The model many companies are moving toward is one I call “virtual vertical integration.” It’s about forming strong partnerships across the value chain — materials, design, foundry, and integration — without trying to do everything under one roof. This approach allows for more focus, better alignment, and a faster path from development to production.
We’ve seen this model reduce time to qualification and support faster design cycles. In a space where demand is growing but resources are limited, this model helps scale without sacrificing performance or quality.
GaN is critical for AI infrastructure
The industry has spent a lot of time talking about compute, but we need to pay more attention to what’s happening at the power level. Without efficient materials, we can’t deliver the infrastructure that AI workloads require. Even if net-zero wasn’t a consideration, we still couldn’t build out the necessary capacity quickly enough using current materials and architectures.
GaN power electronics make AI scale possible. They reduce energy loss in power conversion, support smaller and more efficient designs, and allow more power to be delivered where it’s needed without excessive heat or waste.
Without GaN, the power requirements of AI become unachievable. That’s not an exaggeration. The data center designs we’re seeing today are reaching the limits of what silicon can handle efficiently. GaN extends those limits in a practical, scalable way.
Europe and the West have a window
Recent shifts in geopolitics have highlighted the importance of regional capability. There’s real urgency in the U.S. and Europe to build more domestic semiconductor infrastructure. GaN fits into that picture. It supports energy independence, defense readiness, and digital sovereignty. But to build real capability, you need more than just packaging or design IP — you need control over the materials and epitaxy.
Currently, GaN is at a stage in its adoption curve where leadership can still be established. The window is open. That won’t be the case for long. The countries and companies that invest in GaN supply chains today are setting themselves up to lead in next-generation power, compute, and communications systems.
Looking ahead
The shift toward GaN is happening, and the momentum behind it is real. The performance is there. The cost trajectory is improving. The demand is increasing. But the reliability depends on the materials. That’s where the process starts, and that’s where the foundation has to be solid.
GaN’s transition into mainstream applications is underway. The work now is about scaling with control — getting the materials right, the supply aligned, and the expectations clear across the chain. The companies that treat GaN as a system-level material, not just a device opportunity, are the ones that will move the fastest and build the longest.
The table is set for GaN to move into the mainstream of semiconductors. The industry just needs to stay focused on where the value starts — at the atomic level.
About the author
Dr. Rodney Pelzel is Chief Technology Officer and Chief Operating Officer at IQE, with over 24 years of experience in the semiconductor industry and deep expertise in compound semiconductor materials and epitaxial growth. He has led R&D, operations, and engineering functions at IQE, spearheading numerous new product introductions, including the successful launch of 6-inch VCSELs for consumer applications. A widely recognized thought leader, Dr. Pelzel is the inventor of more than 30 patents and is frequently sought for his insights on semiconductor technology and market trends. He holds a PhD in Chemical Engineering from the University of California, Santa Barbara, and a BS with High Distinction from the University of Colorado. He is a Chartered Engineer, Chartered Scientist, and a Fellow of both the Royal Academy of Engineering and the Institution of Chemical Engineers.
