IN Brief:
- Cohu has received multiple Diamondx orders worth around $5m.
- The systems will test next-generation GaN power devices for AI data centre power architectures.
- Applications include AC-DC conversion, high-voltage DC-DC stages, and point-of-load power management.
Cohu has received multiple orders worth approximately $5m for its Diamondx platform, with the systems set to support development and manufacturing of next-generation gallium nitride power devices for AI data centre power architectures.
The orders, placed by a semiconductor manufacturer, cover Diamondx configurations designed for high-current capability, ultra-low resistance measurement accuracy, and scalable multi-site throughput. The systems will be used for GaN devices deployed across AC-DC front-end conversion, high-voltage DC-DC modules, and advanced point-of-load power management ICs.
Diamondx is Cohu’s air-cooled universal semiconductor test platform, designed to test a wide signal chain spanning power, RF, microcontroller, and mixed-signal devices. The platform scales from five-slot to 40-slot systems and can support more than 500 devices in parallel, depending on configuration.
The system is built for high-throughput production test across high pin-count and high-site-count devices, including wireless components, mobility devices, system-on-chip products, flat-panel display drivers, power management ICs, and microcontrollers. Configured systems can provide more than 7,500 pins, supported by digital, DC, analogue, RF, and automotive instrumentation.
For power applications, Diamondx includes high-density power-management instruments and supports up to 480V testing capability. Its instrumentation covers precision voltage and current sourcing, high-voltage and high-current floating programmable supplies, device power supply channels, and dense voltage/current test functions.
Demand for GaN power devices is being pushed by the rising power density of AI infrastructure. GPU clusters and high-performance AI accelerators require increasingly efficient power conversion, with losses across AC-DC, high-voltage DC-DC, and point-of-load stages directly affecting energy use, heat removal, and rack-level design.
GaN devices can support higher-frequency switching, compact magnetics, and improved efficiency compared with conventional silicon power devices in selected applications. Those advantages are attractive in dense server and data centre power architectures, where power conversion has to shrink physically while handling higher current and tighter thermal limits.
The move toward GaN also raises the bar for production test. Devices need accurate characterisation at high current and low resistance, with switching behaviour, leakage, dynamic parameters, thermal response, and repeatability all feeding into qualification and manufacturing control. As GaN shifts further into volume use, test systems need to combine measurement precision with enough parallelism to keep production cost under control.
AI data centre growth is reshaping more than processor roadmaps. It is increasing demand across power semiconductors, advanced packaging, connectors, thermal systems, rack-level power distribution, and semiconductor test. Higher server power density places conversion efficiency under closer scrutiny, while wide-bandgap devices bring new requirements for validation and production equipment.
Cohu’s Diamondx orders show where capital equipment demand is forming as AI infrastructure moves from compute expansion into the practical engineering of power delivery. The ability to test GaN devices accurately and at scale will sit behind the visible gains in data centre performance, enabling denser power stages without turning manufacturing throughput into the limiting factor.
