GaN wafer production is moving from fixed recipes toward live process judgment
East Asia — South Korea, with global RF, power-electronics and foundry analogues
Share ↗
The signal
Real-time models are beginning to move from after-the-fact inspection into the control loop for compound-semiconductor crystal growth, which could make process data and control software a new source of supplier advantage in yield, uptime and delivery consistency.
The clue
South Korean supplier IVWorks is combining live surface-diffraction measurements and equipment telemetry with automated gallium-nitride wafer growth. This moves process intelligence from inspection after a production run toward intervention while crystals are growing.
Why it matters
Compound-semiconductor suppliers may increasingly compete through proprietary production data and control software—not only material recipes. More stable growth could improve yield, equipment utilization and delivery consistency for communications and power-electronics customers.
Evidence and interpretation
- Observed facts: IVWorks reports that its production system analyzes more than 330 million time-series data points and 30 TB of surface-diffraction data. Its automation handles wafer loading, growth and unloading, while models assess growth conditions. A separate university-industry device using IVWorks’ selective-regrowth process recorded a reported maximum oscillation frequency of 742 GHz at the 2026 VLSI Symposium. The company also raised $4.5 million to expand production infrastructure. - Interpretation: The important pattern is the combination of technical performance, production investment and control software operating on mass-production lines. Conventional operator-led process control and foundry-owned systems remain credible alternatives. - Speculation: If independently verified across customers and multiple lines, accumulated process data could become a defensible advantage for specialty-material suppliers.
What to watch next
The thesis strengthens if independent customer data shows repeatable yield, uptime or qualification gains across multiple production lines; it weakens if the reported improvements remain single-supplier targets or fail to survive production scaling.