IN Brief:
- The Munich District Court has ruled for Infineon in another GaN patent infringement case involving Innoscience.
- Affected products may not be imported, sold, or marketed in Germany, with damages also ordered.
- Repeated patent decisions are adding intellectual-property and continuity risks to GaN component sourcing.
Infineon Technologies has secured another favourable judgment in its gallium nitride patent dispute with Innoscience, with the Munich District Court restricting the import, sale, and marketing in Germany of products covered by its infringement finding.
The court also ordered Innoscience to pay damages. Issued on 3 July, the decision concerns another part of Infineon’s GaN intellectual-property portfolio and follows German judgments delivered in June 2026 and August 2025.
Rather than applying automatically to Innoscience’s complete product range, the injunction covers devices falling within the court’s interpretation of the patent. Product generation, technical design, and regional availability will therefore determine which parts are affected, while any appeal or redesign could alter the commercial position.
Proceedings in Germany run alongside litigation in the United States, where the International Trade Commission found that Innoscience had infringed an Infineon GaN patent and imposed restrictions on affected imports and sales. That earlier ITC decision left the position of redesigned current products contested, with Innoscience maintaining that its continuing commercial range falls outside the restriction.
Infineon holds approximately 450 GaN patent families covering device structures, processing, packaging, and associated power technologies. The company is pursuing several legal actions as GaN moves deeper into data-centre power systems, renewable-energy equipment, industrial automation, communications infrastructure, robotics, chargers, and electric vehicles.
Higher switching speeds and lower switching losses allow GaN devices to support smaller magnetics, higher-frequency conversion, and more compact power stages than many conventional silicon designs. Those gains depend on the complete circuit, however, because gate-drive behaviour, package inductance, thermal resistance, short-circuit response, layout, and electromagnetic compatibility determine how much of the device performance reaches the finished system.
Once a GaN design has completed qualification, a forced component change can extend well beyond replacing one transistor with another nominally similar part. Differences in threshold behaviour, dynamic on-resistance, gate charge, switching transitions, package construction, and thermal paths may require changes to the driver, PCB layout, protection network, cooling system, magnetics, or EMC filtering.
Long-lifecycle industrial and infrastructure products carry the greatest continuity exposure. A power stage expected to remain in production for a decade may depend on a stable device process, controlled manufacturing changes, documented qualification, and access to the same package, while any legal restriction affecting a selected supplier can trigger expensive revalidation.
The dispute is unfolding as semiconductor manufacturers invest heavily in wider-bandgap capacity. Infineon is expanding 300mm GaN production while developing a broader power portfolio spanning silicon, silicon carbide, and GaN, and competing suppliers are building their own processes, package families, and application-support ecosystems.
Larger wafer formats and higher-volume manufacturing should improve cost and consistency, but the economics remain tied closely to proprietary process structures and accumulated manufacturing knowledge. Patent control consequently sits beside yield, package availability, application support, and manufacturing scale in determining which suppliers can compete across high-volume power markets.
Second-source planning remains difficult because GaN products within the same broad voltage and current classes are rarely interchangeable without engineering work. Modular gate-driver design, conservative layout margins, alternate footprints, and early evaluation of more than one architecture can reduce dependence, although each measure consumes board area, development time, or bill-of-materials flexibility.
Procurement controls will also need to identify exactly which devices, revisions, and geographies fall within any restriction. Removing an entire supplier unnecessarily may discard viable qualified components, while assuming that every current part remains unaffected could leave production exposed if a judgment reaches a selected device.
The Munich ruling strengthens Infineon’s position, although the eventual supply effect will depend on appeals, product scope, and Innoscience’s ability to continue supplying redesigned devices. GaN remains a central route towards denser and more efficient power conversion, but electrical performance alone no longer defines the risk attached to component selection.
