Today’s users expect devices that are smarter, faster, more intuitive, more secure, and more private – while remaining affordable. They want seamless intelligence without friction, without thinking about what it takes to deliver it. For product teams, this creates a paradox: delivering that experience requires increasing technical and regulatory effort, which directly impacts development time and cost.
The evolution from IoT to AIoT
In the history of embedded systems, we have moved from the discrete digital logic of the 1960s to the hyper-connected IoT era of the late 2000s, enabled by the ability to connect even small devices reliably to the cloud. This connectivity unlocked entirely new classes of applications, but it also introduced a strong dependency on continuous communication.
Today, we are seeing the next evolution with AIoT (Artificial Intelligence of Things). Advances in hardware and software now make it possible to embed intelligence directly at the edge, allowing devices to operate more autonomously and remain functional even when cloud connectivity is limited or temporarily unavailable. In practice, this leads to hybrid architectures where local intelligence complements the cloud rather than replaces it.
As expectations for ‘smart’ products continue to rise, addressing this paradox within fixed time and budget constraints requires a shift in design philosophy. This shift involves combining the right state-of-the-art technologies to increase engineering efficiency and reduce development complexity.
The rise of the crossover MCU: Bridging the power-performance gap
For years, engineers had to choose between the efficiency of a smaller and more integrated Microcontroller (MCU) and the high-performance complexity of an Application Processor (MPU). The emergence of so-called crossover MCUs has broken this dichotomy. By integrating multiple heterogeneous cores – such as a Cortex-M7 for performance and a Cortex-M4 for real-time tasks – these chips now offer MPU-level power with MCU-level efficiency.
This shift allows us to move away from expensive, power-hungry System-on-Modules (SOM) while retaining the ‘number-crunching’ ability needed for advanced encryption and on-device (artificial) intelligence.
Zephyr: The new standard for embedded portability
As MCUs become more complex, software must become more abstract. The Zephyr RTOS is becoming the “Linux of the embedded world”. Its vendor-independent architecture allows developers to write code that isn’t tied to a specific piece of silicon. Zephyr supports over 900 starter kits, demonstrating its broad adoption across MCU architectures and platforms.
In a landscape where regulatory requirements are becoming mandatory, Zephyr’s built-in security features, from secure boot to encrypted over-the-air (OTA) updates, provide more standardized options for compliance that was previously a custom-build challenge.
The edge AI paradigm: Intelligence without the tether
One of the defining characteristics of next-generation smart products is the ability to process data locally through Edge AI. Rather than relying exclusively on cloud-based intelligence, devices can make decisions closer to where data is generated.
This approach addresses several challenges associated with cloud-dependent IoT architectures, including:
- Latency: Local processing enables faster response times for time-critical actions.
- Privacy: Sensitive data remains on the device, reducing exposure and increasing user trust.
- Energy efficiency: Limiting high-bandwidth data transmission can significantly reduce power consumption.
Recent advances in AI techniques now allow complex neural networks to be optimized so they can run efficiently on cost-effective, low-power hardware.
The compliance catalyst: The role of the Cyber Resilience Act
Meeting high product requirements now includes meeting a new legal reality: the European Cyber Resilience Act. Security can no longer be a ‘patchwork’ effort added late in the cycle; it must be an architectural pillar.
The synergy of this ‘power trio’ provides a strategic path to compliance:
- Hardware-rooted trust: Crossover MCUs provide the physical foundation for secure boot and native encryption.
- Lifecycle management: Zephyr enables the secure OTA updates required to manage vulnerabilities throughout a product’s mandated support window.
- Data sovereignty: Edge AI naturally aligns with privacy-first mandates by reducing the attack surface of the product.
From complexity to synergy
The challenge of modern product development is that sophisticated features often lead to technical complexity, longer development cycles and higher costs. At the same time, products must meet increasingly strict security and regulatory requirements, often invisible to the end user, but critical for market access and long-term viability.
Understanding how the synergy between crossover hardware, scalable OS layers and edge intelligence can reduce engineering effort is key to delivering products that are intuitive and powerful, but also secure and compliant by design.
We are moving toward a future where ‘smart’ is no longer a feature, but an inherent, autonomous quality that evolves over time.