Embedded World 2025 Recap: Top Innovations in Edge AI

The future of AI isn’t in the cloud – it’s in your wrist, your car, and your factory floor, as Embedded World 2025 dramatically showcased.  

Held annually in Nuremberg, Germany, Embedded World is the global hub for embedded, IoT, and in-vehicle system innovation, bringing together top minds in hardware and software. This year’s event carried a theme of scaling AI at the edge across automotive markets and the entire spectrum of IoT.  

For Arm, Embedded World 2025 set the stage for how our ubiquitous compute platform is enabling the ecosystem to bring edge AI to life. Whether through new silicon announcements, groundbreaking AI-enabled software tools, or live demonstrations of real-world applications, the event showcased how Arm-powered innovation is shaping the future of intelligent, efficient computing at the edge.  

Here are the key highlights and takeaways. 

Powering edge AI across all IoT markets

Prior to Embedded World 2025, Arm showcased our commitment to powering the next generation of edge AI IoT applications with the world’s first Armv9 edge AI platform. At the event, Paul Williamson, SVP and GM of the IoT Line of Business at Arm, reflected on this game-changing platform for IoT markets, with it enabling on-device AI models of over one billion parameters – a significant milestone for the industry.  

Learn more about the new Armv9 edge AI platform.

On the Arm Embedded World booth, we showcased Arm’s edge AI capabilities with the generative AI photo booth. Using the YOLO11 model, the Arm-powered Raspberry Pi 5 device detects objects and creates a text story for the user through the TinyStories small language model (SLM). This demonstrates the power of Arm-based edge AI platforms, with all the processing happening on the device. 

Away from the Arm booth, partners from across the IoT and embedded ecosystem highlighted their commitment to using the Arm compute platform across a broad range of AI-powered IoT applications. This was reflected in a flurry of new product announcements at the event. 

Enabling scalable Edge AI processing for industrial IoT

In the industrial IoT space, Renesas introduced the RZ/V2N, the latest addition to its RZ/V Series of MPUs, designed to power the growing high-volume vision AI market. Equipped with four Arm Cortex-A55 CPU cores, an Arm Cortex-M33 core, and an Arm Mali-C55 ISP, the RZ/V2N delivers efficient AI processing for smart factories, robotics, and industrial automation. 

MediaTek unveiled its latest Genio 720 and Genio 520 IoT platforms for high-performance edge AI, which feature Arm Cortex-A78 and Cortex-A55 processors. The platforms are designed for smart home, retail, industrial and commercial IoT applications, supporting generative AI models, human-machine interface, multimedia, and advanced connectivity. 

Infineon announced support for NVIDIA TAO AI models on its PSoC Edge microcontrollers (MCU), which are powered by the Arm Cortex-M55 CPU and Ethos-U55 NPU. This brings advanced AI capabilities a broad spectrum of IoT markets, from low-power embedded systems to high-performance industrial IoT applications used in automation, smart homes, and predictive maintenance.  

Bringing advanced AI capabilities to low-power microcontrollers

Alongside high-performance industrial IoT, there were various partner innovations in the low-power embedded space that were announced at Embedded World 2025. 

Ambiq launched its ultra-low-power AI processor series, the Apollo330 Plus system-on-chip (SoC), which features the Cortex-M55 CPU with Arm Helium technology.  The new series significantly enhances AI performance, while maintaining the ultra-low power consumption. This is a potential game-changer for wearables, smart sensors, and always-on AI IoT applications.  

Meanwhile, Texas Instruments introduced the world’s most compact MCU, the MSPM0C1104. Designed to maximize board space, this ultra-small Arm-based MCU powers highly efficient, space-constrained applications, such as wearables, personal electronics and compact IoT devices, delivering optimal processing performance in a tiny footprint. 

Accelerating software development on Arm 

Following on from the announcement that Arm Kleidi is now available across IoT for performance optimizations and streamlined edge AI development, new software innovations were introduced at Embedded World 2025 that are designed to accelerate the time-to-market for embedded and IoT developers. 

Linaro announced that its ONELab cloud-based testing platform to streamline embedded software development is now commercially available. Built to support to support Arm-based platforms, ONELab enables developers to remotely test, validate, and optimize software across diverse hardware environments, which helps to accelerate time-to-market and improve software quality. Meanwhile, Synopsys introduced Virtualizer Native Execution, a cutting-edge tool designed to run code natively on Arm hardware in the cloud, significantly accelerating chip and software development. 

Advancing automotive intelligence with Arm-powered innovation

Alongside IoT, Embedded World aims to showcase the latest technology innovations across automotive markets, particularly AI-based applications for the software-defined vehicles (SDVs) of the future.  

Just before the event, we announced that Arm Kleidi is now available across automotive markets for accelerated AI. Already leading technology companies are leveraging the performance optimizations and accelerated time-to-market of KleidiAI integrations into the latest versions of popular AI frameworks to boost the development of AI-based applications in the car. One great example was AWS Automotive’s new proof-of-concept AI chatbot for SDVs, with a video featuring Srini Raghavan, Senior Partner Solutions Architect, AWS, describing the demo on the Arm Embedded World booth. 

Learn more on how Arm Kleidi is driving the future of automotive AI. 

Elsewhere at the event, leading automotive partners announced their latest Arm-powered technology solutions for in-vehicle applications. Cipia demonstrated its AI-driven Driver Monitoring Systems (DMS), which runs on Arm Cortex CPUs. This is a real-time, intelligent in-cabin safety solution, which detects driver fatigue, distraction, and unsafe behaviours with minimal power consumption, making it idea for modern vehicles.  

NXP Semiconductors also launched its new Arm-powered S32K5 MCU family, which delivers high performance, power-efficiency, and functional safety capabilities, making the new products ideal for next-generation automotive applications. The S32K5 series also enhances real-time processing and over-the-air updates to help automakers as they develop future SDVs.  

Finally, it wouldn’t be Embedded World without the annual SOAFEE reception. With the initiative now at 150 members, the reception provided the opportunity for attendees to view the latest SOAFEE blueprints for a wide range of in-vehicle applications that are set feature in the SDVs of the future. 

Building edge AI on Arm

Currently, we are at a critical milestone in edge AI, with this scaling across automotive, embedded and IoT markets, and providing advanced performance and power-efficiency. Embedded World 2025 demonstrated the true breadth and capabilities of the Arm compute platform in powering edge AI applications, now and in the future. From the smallest sensor to in-vehicle applications to large industrial IoT applications, Arm technology is everywhere at the edge.