Seco launches industrial edge AI hardware using Qualcomm Dragonwing IQ-2390

This launch provides a standardized path for industrial OEMs to integrate computer vision and real-time processing without the complexity of multi-chip architectures. By consolidating the ISP, NPU, and a RISC-V microcontroller onto a single Qualcomm die, Seco reduces the bill of materials and power requirements for edge devices. Within the broader streaming and vision ecosystem, this move signals a shift toward unified silicon platforms that handle both high-level AI workloads and low-level deterministic control. Watch for the release of Clea OS 3.0 to see how Seco manages software lifecycles and AI model deployment across these distributed industrial nodes.

Seco has been building its industrial edge portfolio around Qualcomm's Dragonwing processor family for over a year. In early 2025, Seco announced its Clea OS platform had surpassed 100 commercial deployments across manufacturing, retail, and logistics verticals, positioning the software layer as a device-management and AI-model-orchestration layer that sits on top of Qualcomm silicon. The Compact Vision 5 and SBC-Dragonwing-IQ-2390 extend that strategy into the entry-level IQ2 tier, targeting cost-sensitive OEMs who previously relied on discrete MCU plus SoC combinations for machine-vision tasks.

Qualcomm has been aggressively courting industrial and embedded OEMs with the Dragonwing brand since rebranding its IoT processor line in late 2024. Qualcomm introduced the Dragonwing IQ-2390 at Embedded World 2025 in Nuremberg, targeting industrial HMI, robotics, and smart-camera applications with an integrated Hexagon NPU rated at 12 TOPS for INT8 inference. The chip also includes a RISC-V based microcontroller subsystem, which Qualcomm positions as a replacement for external MCUs in real-time control loops. Seco is among the first board-level partners to ship product on this specific SKU, alongside SiFive, which provides RISC-V IP cores used in the microcontroller block.

Competing industrial edge AI video analytics platforms from NXP, Texas Instruments, and Renesas are also consolidating NPU and real-time control onto single dies. NXP launched its i.MX 95 family in mid-2025 with an integrated eIQ Neutron NPU delivering 4 TOPS and a Cortex-M7 real-time domain, aimed at similar machine-vision and industrial-control convergence. Texas Instruments responded with the AM62A and AM68A Sitara processors, which pair Arm Cortex-A cores with dedicated deep-learning accelerators for edge vision, though those parts lack the integrated RISC-V microcontroller that Qualcomm and Seco emphasize for deterministic real-time tasks. For OEMs evaluating Seco's new boards, the key differentiator is the single-chip consolidation of ISP, NPU, and RISC-V MCU, which reduces board area and simplifies certification compared to multi-chip designs.