The Industrial Internet of Things (IIoT) is shifting from passive monitoring to the adoption of Physical AI—artificial intelligence systems that interact with the physical world through robotics and autonomous machines. Transitioning to autonomy by 2027 requires restructuring IoT architecture and addressing infrastructure challenges.
Key architectural trends and computing balance
Physical AI requires massive volumes of data, but transmitting all telemetry to the cloud is inefficient. The industry is shifting toward a hybrid two-tier architecture:
- The Edge tier provides primary data filtering, detects anomalies within milliseconds, and triggers local response scenarios.
- The Cloud tier aggregates data for global analytics and long-term storage.
Standardization and operational technology security
Data unification is critical for model training. The OPC UA standard enables the normalization of heterogeneous data from multi-vendor sensors into a unified format before analysis.
In operational technology (OT), according to the NIST SP 800-82 guidelines, operational availability is the top priority. To secure infrastructure, the following measures are being implemented:
- Strict network segmentation to isolate critical equipment from the corporate IT segment.
- Application of ISA/IEC 62443 standards for industrial automation.
- Creation of demilitarized zones (DMZs) for data exchange between the Edge tier and the cloud.
Lifecycle management and secure deployment
Scaling edge nodes requires a systematic approach. According to AWS Well-Architected IoT Lens practices, the device lifecycle must include secure onboarding, regular firmware updates, and continuous anomaly monitoring.
Experts from Softengi (a member of the Intecracy Group alliance) note that designing Physical AI systems in compliance with the ISO/IEC 42001:2023 standard enables the creation of resilient infrastructures while meeting stringent cybersecurity requirements.
What this means for the market
The shift to Physical AI means businesses can no longer rely on traditional cloud-only IoT frameworks. Companies must rapidly adapt to hybrid architectures to prevent bandwidth bottlenecks, while strict compliance with industrial security standards becomes a baseline requirement for maintaining operational availability.
What to do next
- Restructure architecture: Deploy a two-tier Edge/Cloud system to balance data processing and reduce latency.
- Standardize data: Adopt the OPC UA standard to normalize heterogeneous data from multi-vendor sensors.
- Secure the infrastructure: Implement network segmentation, DMZs, and align deployments with ISA/IEC 62443, NIST SP 800-82, and ISO/IEC 42001:2023 standards.
Prepared by a Software Ukraine member. Original publication.