The gap between an AI prototype and a production-ready industrial machine can be wider than the prototype itself. SECO is targeting that transition with early access to its SOM-SMARC-Dragonwing-IQ8, a compact system-on-module based on Qualcomm Technologies’ Dragonwing IQ8 Series and designed to carry applications developed with Arduino into industrial Edge AI deployments.
Prototyping an AI-enabled robot or industrial vision system is becoming easier. Turning that prototype into a product that can operate reliably for years is still a considerably harder engineering problem.
SECO is attempting to narrow that gap with priority access to early samples of its SOM-SMARC-Dragonwing-IQ8, a system-on-module aimed at OEMs developing industrial Edge AI equipment.
The module is based on the Qualcomm Dragonwing IQ-8275 processor and conforms to the SMARC 2.1.1 embedded computing standard. SECO developed the platform through its collaboration with Arduino and Qualcomm Technologies, with applications including industrial robotics, smart machines, machine vision, human-machine interfaces and machine control.
The bigger story, however, is less about another embedded AI module than about the transition from development board to industrial product.
Arduino has become a popular environment for rapidly validating hardware and software concepts. Industrial OEMs face a different set of requirements once an idea moves toward commercialization: lifecycle availability, cybersecurity, thermal and power constraints, software maintenance, remote management and production-scale deployment all become important.
SECO’s proposition is that engineers should not have to start over at that point.
Preserving the prototype investment
The SOM-SMARC-Dragonwing-IQ8 is designed to work with Arduino VENTUNO Q, an Edge AI development platform built around the Dragonwing IQ8 Series.
Developers can prototype and validate applications on the Arduino platform and use Arduino App Lab to help port those applications toward SECO’s industrial system-on-module.
That continuity could be valuable for engineering organizations.
A prototype often contains more than an AI model. It may include application logic, device interfaces, computer-vision pipelines and other software work accumulated during weeks or months of development. Rebuilding those components for production hardware can add substantial engineering effort and introduce new integration risks.
The industrialization challenge is therefore partly about hardware and partly about preserving the software development path.
SECO adds its own industrial capabilities around that foundation, including board design expertise, board support package and operating-system support, cybersecurity services and lifecycle management.
The company also provides support for the Clea framework, extending the embedded platform into connected-device management.
Edge AI moves closer to the machine
The module is designed for applications where AI processing needs to happen near the equipment generating the data.
That matters in industrial environments because sending every camera frame or sensor reading to a cloud platform can introduce latency, connectivity dependencies and additional data-management considerations.
On-device or edge processing can allow systems to make decisions locally.
For a machine-vision application, that might mean analyzing an image at the production line rather than sending it to a remote server. For robotics, local compute can support responsive perception and control. For industrial monitoring, it can enable processing even when network connectivity is limited.
The SOM-SMARC-Dragonwing-IQ8 offers configurations reaching up to 40 TOPS of AI compute, according to SECO.
The platform supports up to 32GB of LPDDR5/LPDDR5X memory and up to 1TB of UFS 3.1 storage. Connectivity includes Gigabit and 2.5 Gigabit Ethernet, PCIe Gen4, MIPI-CSI and CAN-FD.
Those specifications put the module into a category where developers are balancing AI performance against power consumption, physical size and bill-of-materials costs rather than simply maximizing compute.
Qualcomm enters through the industrialization layer
The collaboration also reflects Qualcomm’s broader expansion beyond smartphones into industrial IoT, robotics and edge computing.
The Dragonwing portfolio is aimed at connected and intelligent devices that require dedicated compute at the edge. Qualcomm’s competitive field includes embedded processors and AI platforms from NVIDIA, AMD, Intel, NXP and other silicon vendors, as well as specialized edge-AI accelerators.
For OEMs, silicon performance is only one part of the decision.
The ability to obtain production-ready modules, software support, security updates, carrier-board expertise and long-term lifecycle management can matter just as much when a design enters a multi-year industrial deployment.
That is where SECO is attempting to differentiate its offering.
Clea adds the fleet-management layer
The hardware is paired with support for SECO’s Clea framework, which provides capabilities including fleet management, secure remote OTA updates, remote troubleshooting and container orchestration.
This is increasingly important as industrial AI deployments move beyond individual machines.
An OEM may eventually need to manage hundreds or thousands of devices distributed across factories or customer sites. Updating models, patching software, diagnosing failures and controlling deployments remotely becomes an operational requirement rather than an optional feature.
Clea Studio AI is also positioned by SECO as a way to accelerate AI workflow deployment at scale.
In that sense, the platform extends beyond an embedded computing component. The intended stack runs from development and AI application creation through hardware deployment and fleet operations.
Arduino’s ecosystem becomes part of the pitch
The “Works with Arduino” positioning is another important component.
Arduino’s strength has historically been its developer accessibility and ecosystem. Industrial hardware, by contrast, tends to prioritize robustness, lifecycle and deployment requirements.
The combination attempts to bridge those two worlds.
Marcello Majonchi, chief product officer at Arduino, described the goal as maintaining continuity between the idea stage and industrial production. Qualcomm similarly emphasized a common hardware and software foundation as a way to reduce development effort and accelerate time to market.
Whether that translates into significant savings will depend on individual applications and how much existing prototype software can be reused. But the underlying industry problem is real: moving from a development platform to production hardware frequently introduces new dependencies and engineering work.
The industrial Edge AI market is becoming a platform race
The announcement arrives as industrial companies increasingly experiment with AI-powered inspection, robotics, predictive maintenance and intelligent automation.
The market is consequently moving beyond the question of whether edge AI can run a model. OEMs increasingly need a complete architecture encompassing compute, connectivity, software, security, device management and lifecycle support.
SECO’s approach addresses that broader stack.
The SOM-SMARC-Dragonwing-IQ8 does not attempt to compete solely on raw AI performance. Its proposition is a more integrated route from Arduino prototyping to production-oriented industrial Edge AI, backed by Qualcomm silicon and SECO’s embedded engineering and lifecycle services.
For manufacturers and OEMs, that could be the more consequential metric.
The winning edge-AI platform may not simply be the one with the highest TOPS figure. It may be the one that lets an engineering team turn a validated AI concept into thousands of maintainable machines without discarding the work that made the original prototype successful.
Market Landscape
Industrial Edge AI is becoming increasingly competitive as NVIDIA, Qualcomm, AMD, Intel, NXP and other semiconductor companies push AI compute into robots, cameras, vehicles and industrial equipment.
The market is shifting from standalone accelerators toward complete platforms combining compute, software, connectivity, security and device management.
SECO’s strategy occupies an interesting position in that market: rather than targeting developers exclusively with a development kit, it focuses on the prototype-to-production transition.
For enterprise OEMs, the key evaluation criteria will include AI performance, power efficiency, software portability, ecosystem support, cybersecurity, lifecycle commitments and total development cost.
The Arduino connection could lower the barrier to experimentation, while Qualcomm’s Dragonwing architecture provides the underlying compute platform. SECO’s industrial services and Clea address the deployment and lifecycle side.
Top Insights
- SECO is opening early access to a Dragonwing IQ8-based SMARC module, giving industrial OEMs a path from Arduino Edge AI prototypes toward production-oriented robotics and machine systems.
- The platform targets up to 40 TOPS of AI performance, alongside LPDDR5 memory, UFS storage and industrial connectivity for compute-intensive edge applications.
- Arduino VENTUNO Q and SECO’s SOM share a development path, potentially reducing software redevelopment when engineering teams move validated prototypes into industrial hardware.
- Clea adds fleet-management capabilities, including secure OTA updates, remote troubleshooting and container orchestration for organizations deploying AI-enabled machines at scale.
- The competitive edge may be ecosystem continuity rather than compute alone, as OEMs increasingly evaluate industrial AI platforms on lifecycle support, security and deployment economics.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
