At embedded world 2025, Tria Technologies unveiled a new family of modules built around Qualcomm’s Dragonwing processors. The modules support multiple operating systems, including Yocto Linux, positioning them as a flexible option for embedded Linux developers.
By combining high-performance, low-power processors with Linux support, Tria aims to enable advanced Edge AI, machine learning, and IoT applications across sectors such as industrial, medical, agriculture, and construction.
To explore this further, we spoke with Christian Bauer, Product Marketing Manager at Tria Technologies, about the technology behind the modules, the design challenges they addressed, and how the company envisioned the roadmap ahead.
Tria Technologies has the largest SMARC and standard module portfolio in the market. What strategic role does these five new products play in consolidating or expanding that leadership?
Tria Technologies’ launch of five new compute modules – IQ‑615, QCS5430/QCS6490, IQ‑9075 Vision AI‑KIT, and Snapdragon X Elite COM Express/COM-HPC – is a decisive move to consolidate and expand our leadership in the OSM, SMARC, and COM Express/HPC markets. By partnering with Qualcomm Technologies as a leader in the market, we are able to leverage their latest technologies to develop solutions which deliver exceptional computing power and efficiency for a large range of embedded and Edge AI applications. This not only enables us to remain at the forefront of the industry as the largest SMARC portfolio provider but also encourages further trust with our customers that we will continue to develop and provide innovative solutions with the latest technologies available to meet the demands of a fast-paced market. With the latest IoT product from Qualcomm Technologies we are able to deliver highest ARM computing- and AI-power on OSM, SMARC, and COM Express/HPC.
We feel that by aligning each product with specific workload and thermal profiles, Tria empowers engineers to scale across use cases without leaving the OSM, SMARC, or COM- Express/HPC environments. All modules support Linux and select models are now certified for Android16 and Windows on ARM, which enables developers to meet diverse OS and application requirements. Backed by our engineering expertise, long-term availability, and optional carrier board customisation, this expanded portfolio enhances time-to-market, futureproofs designs, and ensures system resilience across industrial-grade deployments. The result is a more versatile, performance-tiered standard module strategy that solidifies Tria’s foothold in emerging Edge AI markets while opening new paths to growth.
What market segments are you specifically targeting with these new modules – are there particular verticals where you see immediate traction?
At their core, these new modules provide advanced AI acceleration, low power consumption, and broad OS compatibility – including Linux and Windows on ARM – which make any of them an ideal choice for a wide range of Edge applications. Some of the applications they are best suited for range from industrial automation, robotics, CAV (commercial and agricultural vehicles) to healthcare and smart mobility, with each family optimised to meet the performance, connectivity, and longevity needs of its specific market.
This product diversity is important because Tria customers range across a substantial number of application areas, so our focus is on ensuring Tria can support all of them with the same level of product and service excellence. That said, Tria places special emphasis on vertical segments, which includes industrial automation and robotics, commercial and agricultural vehicles, plus high-growth medical, prosumer, and building automation applications and these new modules are a perfect fit for their respective needs.
What makes Tria’s new modules uniquely suited for deployment in IoT devices and AI applications?
Tria’s new module family, based on Dragonwing platforms, enables us to address virtually any requirement, while conserving power, upgrading performance, and delivering AI capabilities for use in challenging environments, across a wide range of module standards and applications.
And we are equally excited as we see our new modules improve AI performance, which in turn opens vast new markets for AI deployment at the Edge. All our new modules have an in-built AI accelerator, and each is designed to operate at low, energy-saving power levels with no compromise in performance.
The product families themselves are positioned in logical groupings, such as:
- Entry-level modules (on OSM and SMARC) focus on ultra-low power Edge nodes ideal for agriculture and remote monitoring systems where efficiency and size are critical criteria
- Mid-tier families (on SMARC) support more complex workloads in, for example, medical devices and industrial control systems to achieve the best balance of performance and power
- High-performance variants (on COM Express/COM-HPC and 3,5” SBC) feature Snapdragon X Elite and IQ9, are aimed at compute-intensive Edge AI and robotics applications
How does the German design and manufacturing aspect enhance reliability, supply chain confidence, or product performance?
We are proud of the fact that most of Tria’s manufacturing takes place in Germany and Italy because it makes us highly visible and accessible throughout Europe. The ability to share our expertise with those who need it, and do so quickly, is seen as a significant advantage for Tria customers and is a major differentiator for us in the market. And because Tria uses its own, local facilities, we can maintain a strong focus on delivering high quality and competitive pricing.
It is important to recognise that a reputation is not built solely on delivering great hardware, software and security support are key aspects too. You cannot just ship a module off to a customer and forget about it. The explosive growth of AI and its applications means that support is more important that it has ever been.
But with that growth comes oversight. Tria has established a task force that examines local, regional, and international regulations and makes sure our products align with current, and foreseen, legislation. We also have a strong software team that helps customers during the design process, with both the intricacies of software integration and the regulated aspects of every project to ensure the design is compliant.
Tria started module development and production many years ago with strong R&D and manufacturing in Germany. We have now added teams in the Americas, Italy, and China to create a global design network. We are also continuing to expand our R&D capabilities in 2025.
German design and manufacturing enhance product reliability and performance through precision engineering, strict quality control, and close integration between design and production. These strengths are amplified by Avnet’s global distribution network, which provides robust supply chain support, reduced lead times, and secure inventory management. With Avnet’s extensive logistics infrastructure and customer reach, products designed and manufactured in Germany can be delivered efficiently and consistently worldwide, ensuring customers benefit from both high-quality engineering and dependable supply chain execution.
Tria’s new modules now support Android 16, Windows 11 IoT Enterprise, and Yocto Linux. How important is this multi-OS capability to developers working across vertical markets?
Multi-OS support is increasingly critical for developers working across diverse verticals, and our decision to enable Android 16, Windows 11 IoT Enterprise, and Yocto Linux across our latest standard modules reflects a deep understanding of these varied requirements. Another example is in industrial automation, where developers tend to prefer Yocto Linux for its customisation and deterministic behaviour in real-time control systems. At the same time, Windows 11 IoT Enterprise – now supported on Tria’s Snapdragon X Elite and QCS5430/6490 modules – offers a familiar development environment and long-term support for machine builders and integrators developing HMI applications, especially in factories and logistics hubs where existing Windows infrastructures are prevalent.
In sectors like smart retail, smart cities, and consumer-facing IoT, Android is frequently the OS of choice due to its mature ecosystem, rapid UI development, and native support for camera and touchscreen interfaces. Tria’s new modules, such as the QCS6490/QCS5430, are well-suited for Android-based kiosks, point-of-sale terminals, and digital signage solutions that rely on intuitive, high-performance user experiences. By offering these OS options across its scalable hardware – from ultra-low-power Edge nodes to high-performance AI platforms like our IQ‑9 Vision AI‑KIT – Tria enables customers to optimise their software stacks without being constrained by hardware limitations. This flexibility accelerates development and integration and also reduces total cost of ownership.
Are you seeing early demand for Windows on ARM in particular industries or regions?
Yes, and that is why Tria also now provides Microsoft Windows on ARM. In fact, we have been surprised at the overwhelmingly positive market reaction to Windows on ARM, especially customers with HMI applications, or customers coming from x86. The ability to enable a seamless transition from the x86 architecture to ARM on the same OS had not been possible until Windows on ARM became available, and we are seeing the demand grow apace. It is fair to say that the momentum behind the growth of Windows on ARM has been steady rather than explosive. Windows was, after all, designed to use x86 architecture, but as Windows on ARM becomes more popular, there are no barriers to using it for those who want the longer battery life and improved performance in certain areas that ARM delivers.
Are there plans to expand the Qualcomm-powered range beyond these initial five modules?
One of the many benefits of our collaboration with Qualcomm is that it enables us to respond quickly to evolving market demands across industrial, retail, medical, and smart infrastructure sectors. The five platforms we launched at embedded world 2025 are just the beginning. We fully intend to grow our portfolio with new offerings as Qualcomm introduces next-generation platforms. Our role is to ensure these technologies are production-ready and easy to adopt – backed by flexible OS support, industrial-grade reliability, and long-term availability. That is how we help customers get to market faster, with the right performance point for every application.
As more embedded compute designs incorporate OS flexibility and AI capabilities, what future enhancements or product roadmap developments can we expect from Tria?
We always carefully review our suppliers’ roadmaps for selecting next-generation processors that best fit our customers’ needs. This includes criteria such as performance, IO capabilities, support for our industrial temperature range, power dissipation, and longevity.
We will be working with our partners such as Qualcomm, to continue to be heavily invested into developing solutions with the latest technologies to meet market demand, all with the driving goal to ensure value for our customers.
Regardless of the requirement, enhancing security, streamlined software development, intuitive software frameworks, an optimised build environment, or pre-compiled models for AI applications, Tria is well positioned to deliver comprehensive solutions.
How do you see the role of Edge AI evolving over the next two to three years, and how is Tria preparing for that shift?
Edge AI is rapidly becoming a cornerstone of innovation – from industrial automation and agriculture to security and quality control. As AI use cases grow more diverse and demanding, the need for real-time, on-device intelligence is similarly rising. This shift in emphasis is driven by requirements for low latency, data privacy, and reliable operation in environments where cloud connectivity is limited or impractical.
Tria Technologies and Qualcomm are preparing for this transformation by delivering highly flexible, standardised compute modules powered by Snapdragon and Dragonwing SoCs that support a wide range of AI workloads – from lightweight models for voice and anomaly detection to high-performance video analytics with up to 100 TOPs of compute.
We also leverage advanced Edge AI techniques like quantisation, layer fusing, and model pruning to maximise efficiency while maintaining accuracy. By giving developers access to scalable hardware, pre-trained models, and robust development tools, Tria is enabling faster time to market and making Edge AI more accessible and impactful across embedded applications.
