With embedded design getting more complex, Grinn seems to rewrite the rulebook, making them simpler and smarter. In an exclusive chat, Robert Otreba and Kateryna Kozakova reveal to EFY’s Akanksha Sondhi Gaur how AI-ready SOMs and a forward-looking MediaTek partnership are powering the future of edge intelligence.
Q. Could you briefly introduce the company’s journey, its origin, and core mission today?
A. Grinn began in Poland as a small team focused on embedded systems and Internet of Things (IoT) innovation, and has grown over nearly two decades into a company of about 70 engineers supporting global customers in building sophisticated, artificial intelligence (AI)-enabled products. Its mission remains to simplify embedded development by providing high-performance system-on-modules (SOMs) that integrate compute, connectivity, sensors, and AI in compact, ready-to-deploy platforms, supported by end-to-end design expertise so innovators can focus on creating great products.
Q. What differentiates you in the highly competitive embedded systems and IoT ecosystem?
A. It is the depth and breadth of our engineering capabilities in wireless connectivity, edge computing, and complex embedded design. Unlike companies that offer only hardware or software, we deliver fully integrated solutions with extremely compact SOMs that provide strong performance, power efficiency, and inherent design flexibility, enabling original equipment manufacturers (OEMs) to accelerate time-to-market and adopt modern processors without overwhelming complexity.
Q. What drives the value and focus of your SOM-based approach?
A. Our SOMs were created ‘by engineers, for engineers’ to remove the toughest barriers in embedded development. As processors integrate central processing units (CPUs), neural processing units (NPUs), memory, and complex power architectures, SOMs package this complexity into compact, pre-validated, AI-ready compute modules supported by documentation, development kits, reference designs, software development kits (SDKs), and sample applications, enabling faster prototyping, reduced risk, shorter time-to-market, and more reliable designs.
Q. What is your main unique selling proposition (USP) in the SOM market?
A. Our primary USP is delivering some of the smallest SOMs in their performance category, without compromising computing power or power efficiency, while ensuring flexibility, component longevity, and compatibility across variants, critical for industrial markets requiring reliability and long lifecycle stability.
Q. What trends shape your research and development (R&D) roadmap?
A.Edge AI, increased use of NPUs, and demand for real-time vision and robotics drive our roadmap. Industries such as industrial automation, smart agriculture, and smart homes require intelligent embedded platforms; therefore, every new SOM we design is AI-capable and future-proof in connectivity. We invest ahead of market trends to help customers scale early with confidence.
Q. How do you build innovation within engineering teams?
A. nnovation extends beyond SOMs to the entire ecosystem, with detailed SDKs, development kits, reference hardware, whitepapers, AI demos, and software examples that guide customers from proof of concept to production, helping them build faster and reduce risk.
Q. What is the origin and impact of your MediaTek partnership?
A. Our partnership with MediaTek is based on aligning with a semiconductor company offering powerful yet energy-efficient processors, industrial reliability, and strong AI capabilities. Early chipset access enables co-development, ensuring our SOMs deliver high computational performance, power efficiency, multimedia capability, and robust AI for advanced edge applications such as analytics, automation, and robotics.
Q. What MediaTek SOMs are in development, and how do you tune integration?
A. We offer SOMs based on MediaTek’s Genio 510 and Genio 700, designed to be pin-to-pin compatible, and are developing another SOM to broaden our performance range further. A dedicated engineering team optimises hardware, software stacks, drivers, SDKs, reference implementations, and AI enablement to ensure fully optimised, production-ready platforms.
Q. How does this partnership reduce design risk and accelerate time-to-market?
A. MediaTek processors are powerful but complex. Our SOMs provide pre-validated, pre-certified compute modules that eliminate the need to design high-speed memory, power domains, or radio frequency (RF) subsystems from scratch, dramatically reducing risk and potentially shortening development cycles by up to a year.
Q. How has the partnership shaped your R&D culture?
A. It has accelerated our R&D cycle by providing early access to reference drivers, documentation, and next-generation chipsets, enabling proactive platform development. Our workflow spans research, feasibility, proof of concept, prototyping, hardware and software design, AI integration, validation, certification, and manufacturing support, with customers able to engage at any stage.
Q. How do you align hardware and software and balance custom versus standard SOMs?
A. Our hardware and software teams work in parallel to optimise thermal performance, power management, I/O throughput, and AI efficiency, ensuring cohesive platforms. Standardisation is balanced with flexibility through configurable memory, storage, and temperature options, with deeper customisation at the carrier-board level.
Q. How does your SOM design process reduce customer risk?
A. We handle high-speed processing interfaces, power architectures, electromagnetic compatibility (EMC), and thermal robustness upfront. Customers can begin prototyping using evaluation kits and rely on our schematics and layout guidelines during production.
Q. How do you ensure lifecycle reliability and component strategy?
A. We engineer SOMs for long-term reliability, validating every unit and selecting components with 10–12 years availability, industrial-grade robustness, and stable supply to avoid redesigns and support long-term deployments.
Q. What is the toughest SOM design challenge?
A. Hidden failures often arise from power-domain interactions under thermal stress, while the toughest challenge remains preserving signal integrity in compact footprints alongside power, DDR memory, NPUs, thermal control, and electromagnetic interference (EMI) management.
Q. Can you share a moment where your SOMs solved an ‘impossible’ requirement?
A. A customer building a compact, low-power device requiring real-time computer vision reduced expected processor and memory integration time from more than a year to a working prototype in weeks using our SOM.
Q. Any recent demos showcasing engineering capability?
A. We recently showcased a robotic arm with real-time vision analytics at MediaTek’s booth at Embedded World and are developing additional robotics, computer vision, and edge AI demos.
Q. How do your in-house labs accelerate development?
A. Our labs feature high-speed oscilloscopes, RF instrumentation, thermal chambers, EMC pre-compliance tools, and reliability rigs to validate SOMs and support customers, reducing late-stage failure risk and shortening development cycles.
Q. What support do you offer design engineers?
A. We provide documentation, a GitHub repository, SDKs, reference designs, evaluation kits, single-board computers, and engineering guidance for carrier boards, drivers, and AI integration to ensure a smooth transition from concept to production.
Q. How do you maintain developer friendliness amid chipset complexity?
A. Developer experience is treated as a core feature, supported by documentation, clean application programming interfaces (APIs), tutorials, AI demos, and debugging guidance.
Q. What architectural changes will SOMs need as AI moves to the edge?
A. SOMs will shift toward AI-centric architecture with more powerful NPUs, heterogeneous compute, AI-optimised memory, multi-sensor synchronisation, advanced security engines, and integrated toolchains for model compression and retraining.
Q. What innovations and long-term AI vision are you pursuing?
A. We are expanding upward and downward with both cost-efficient AI modules and higher-performance platforms, while investing in tools, demos, and documentation. Long-term, we aim to be a globally trusted provider of AI-first SOMs with compact design, strong compute, lifecycle stability, and a robust developer ecosystem.
Q. Which global regions are a priority?
A. We are expanding in Europe, the United States, and Asia. MediaTek is helping deepen our presence in the Asian market.
Q. How do you ensure localisation without lowering standards?
A. We work with distributors and integrators while maintaining global consistency through CE and Federal Communications Commission (FCC) pre-certified SOMs.
Q. How do you view India for SOMs and local collaboration?
A. India is a rapidly growing IoT and industrial electronics market with strong engineering talent suited to our SOM strengths. Opportunities lie in smart cities, industrial IoT and Industry 5.0, robotics, computer vision, smart agriculture, and intelligent gateways. While we are not planning a physical office yet, we are building strong local networks and are open to partnerships with Indian design houses, startups, and academia.
