Ezurio Carbon AM67 SOM delivers 4K60 video and 4 TOPS AI

The launch of this module provides a high-density solution for streaming and vision tasks that require local processing without cloud latency. By combining 4K60 hardware acceleration with dedicated AI silicon, Ezurio is targeting the growing demand for sophisticated image signal processing in ruggedized environments. This move strengthens the Texas Instruments ecosystem by providing a certified, production-ready path for the AM67x processor family. As edge devices increasingly handle complex video analytics and multi-display outputs, this hardware reduces the integration burden for OEMs. Watch for the start of mass production in October to see how quickly industrial vision providers adopt this OSM-MF form factor.

Ezurio's parent company, Laird Connectivity, rebranded to Ezurio in 2023 and has since expanded its system-on-module portfolio to target edge AI and industrial vision markets. The Carbon AM67 sits within a broader ecosystem of TI AM67x-based designs that Texas Instruments has positioned for industrial edge AI inference at under 5 watts of power consumption, competing against modules from vendors like Phytec, Variscite, and iEi Integration that also target the OSM form factor. The OSM (Open Standard Module) specification, maintained by the SGET consortium, defines standardized pinouts and mechanical dimensions to reduce carrier-board redesign costs, and the OSM-MF variant used by the Carbon AM67 measures just 22mm x 25mm, making it one of the smallest footprints available for 4K-capable video processing.

The competitive landscape for edge AI modules with integrated video acceleration has intensified in 2026. NVIDIA's Jetson Orin Nano 2, announced in late 2024, delivers 67 TOPS of AI performance at a $249 price point, setting a high benchmark for AI throughput per dollar that TI-based modules must differentiate against through power efficiency, video codec support, and industrial temperature ratings rather than raw inference speed. Meanwhile, Qualcomm has expanded its QCS series of edge AI processors into industrial HMI and vision applications, offering integrated Wi-Fi and Bluetooth connectivity alongside AI acceleration, a combination that Ezurio addresses through its companion Sona TI351 and CC3351 wireless modules rather than on the SoC itself.

Technical differentiation for the Carbon AM67 centers on its dedicated video codec hardware and multi-display support. The AM67x processor includes a hardware video codec accelerator supporting H.264 and H.265 encode and decode at 4K60, which Texas Instruments documented in its AM67x technical reference as capable of simultaneous encode and decode streams. This dual-stream capability is relevant for edge vision systems that must ingest one camera feed while outputting a processed or annotated stream to a display. The 4 TOPS of AI performance comes from TI's C7x DSP and MMA (Matrix Multiply Accelerator), which TI has benchmarked at 3.2 TOPS per watt in its published efficiency comparisons against competing edge processors. For streaming and broadcast applications, the combination of hardware codec acceleration with low-power AI inference positions the Carbon AM67 for use cases like automated camera switching, real-time graphics overlay, and on-device content analysis where cloud round-trip latency is unacceptable.