AI Is Moving Beyond Software Into the Industrial World
NVIDIA GTC Berlin 2026 highlighted a significant shift in the relationship between artificial intelligence and industrial technology. Rather than focusing exclusively on GPUs, generative AI models, or computing performance, the event presented a broader vision in which AI becomes an integral part of engineering, manufacturing, robotics, telecommunications, and industrial infrastructure.
The central development is the emergence of an integrated industrial AI architecture. Foundation models, digital twins, simulation platforms, edge computing, autonomous systems, and AI factories are increasingly being developed as interconnected components rather than independent technologies.
For industrial companies, this transition could fundamentally change how production systems are designed, operated, maintained, and optimized. AI is no longer simply an additional analytical tool. It is gradually becoming part of the operational architecture itself.
Europe Has an Opportunity to Shape the Global AI Industry
A major message from NVIDIA CEO Jensen Huang's keynote was that Europe should play a more active role in building the global AI economy. The region already has substantial industrial capabilities in manufacturing, automotive engineering, energy infrastructure, telecommunications, and automation.
These established industries provide a strong foundation for applying AI to real-world industrial challenges.
Several European telecommunications operators, including Deutsche Telekom, Orange, Elisa, and BT, are developing AI-RAN infrastructure and distributed AI capabilities. Automotive manufacturers such as Mercedes-Benz, Volvo Cars, and Volkswagen Commercial Vehicles are exploring NVIDIA-based platforms for autonomous mobility. Meanwhile, industrial technology companies including ABB, KUKA, Schaeffler, Comau, and SICK are integrating AI technologies into robotics, simulation, and engineering workflows.
In my view, Europe's competitive advantage will not come from competing with other regions solely on GPU production or cloud computing capacity. Its greater opportunity lies in combining existing industrial expertise with AI capabilities.
A manufacturer that understands production processes, equipment behavior, process control, and industrial safety has valuable knowledge that cannot be replaced simply by deploying a larger AI model. The companies that successfully connect this operational knowledge with AI will be better positioned to develop practical industrial applications.
AI Factories Are Becoming a New Category of Industrial Infrastructure
Another important development at GTC Berlin was NVIDIA's continued expansion of its AI Factory concept. The introduction of DSX POD reflects a shift toward treating AI infrastructure as an engineered production system rather than a collection of independently installed computing components.
Traditional AI data center projects require extensive integration of computing hardware, networking, electrical distribution, cooling, and storage systems. These components must be configured and tested together before the facility can operate at its intended capacity.
DSX POD introduces a more modular approach. Infrastructure can be assembled, integrated, and validated in a factory environment before being delivered for deployment. This reduces the amount of complex integration work required at the installation site.
The approach has clear similarities to established industrial engineering practices, including modular construction, prefabrication, standardized equipment packages, and factory acceptance testing (FAT).
From an automation engineering perspective, this is an important development. Large AI facilities have many of the same engineering challenges found in industrial plants: power distribution, thermal management, equipment monitoring, communication networks, system coordination, and operational availability.
Standardizing these infrastructure elements could make AI capacity easier to expand and maintain. It also creates opportunities for industrial automation suppliers to contribute through power management, intelligent monitoring, control systems, and integrated facility management.
Schneider Electric Connects AI Computing With Power and Cooling Engineering
Schneider Electric's announcement of its RapidScale Modular Data Center Reference Design, built around NVIDIA Vera Rubin infrastructure and advanced liquid cooling, illustrates how traditional industrial infrastructure technologies are becoming increasingly important to AI development.
As AI computing systems become more demanding, electrical power delivery and heat removal are no longer secondary facility considerations. They directly influence computing density, system performance, operating costs, and infrastructure scalability.
Liquid cooling is particularly relevant because high-density AI computing environments generate substantial thermal loads. Managing these loads requires coordinated engineering across cooling equipment, electrical systems, monitoring devices, and facility control infrastructure.
This creates a broader role for industrial automation technologies. Sensors, controllers, power monitoring equipment, communication networks, and supervisory systems can help operators maintain visibility over critical infrastructure conditions.
One point deserves particular attention: AI factory performance should not be measured only by computing capacity. Power efficiency, cooling effectiveness, equipment utilization, maintainability, and system availability are equally important operational indicators.
For industrial operators, the real challenge is to ensure that the infrastructure supporting AI remains manageable as computing requirements continue to increase.
Digital Twins and Simulation Are Becoming Core Engineering Tools
The convergence of AI and industrial engineering also places digital twins and simulation technologies in a more central position.
Traditional industrial engineering often relies on physical prototypes, commissioning activities, and field testing to verify system behavior. These processes remain necessary, but they can be expensive and time-consuming, particularly when dealing with complex robotic systems and highly integrated production environments.
Digital twins provide a virtual representation of physical assets, processes, or entire facilities. When combined with simulation and AI, they allow engineers to investigate operating conditions, test alternative configurations, and evaluate system behavior before making changes to real equipment.
For example, a manufacturer could use simulation to evaluate robot movements, production-line layouts, material handling sequences, or interactions between machines. Potential problems can be identified earlier in the engineering process, reducing the need for repeated physical adjustments.
However, a digital twin is only as useful as the information supporting it. Accurate equipment models, realistic operating parameters, reliable sensor data, and consistent engineering documentation are essential.
My assessment is that industrial companies should avoid treating digital twins as visualization projects. Their real value comes when simulation results can support engineering decisions, commissioning, operational optimization, and maintenance planning.
Physical AI Is Bringing Intelligence Closer to Machines
One of the broader implications of NVIDIA's industrial strategy is the growing importance of Physical AI. Unlike conventional AI applications that primarily process text, images, or business data, Physical AI is designed to interact with the physical environment through robots, autonomous machines, and intelligent equipment.
This requires a combination of perception, reasoning, simulation, motion planning, and real-time interaction with physical systems.
Industrial robotics is a natural application area. Robots must recognize objects, understand their surroundings, respond to changing conditions, and execute movements within defined safety and operational limits.
The integration of AI with industrial robotics could expand automation beyond highly repetitive and predefined tasks. It may enable machines to handle more variable production environments and adapt to changing operational requirements.
Nevertheless, intelligent behavior does not automatically mean industrial readiness. Robot motion, emergency stops, protective functions, machine interlocks, and safety-related control systems must remain subject to appropriate engineering validation.
AI can improve the flexibility of automation, but deterministic control and established safety architectures will continue to play a fundamental role in industrial applications.
The Industrial AI Stack Requires More Than Computing Power
The most important takeaway from GTC Berlin 2026 is the emergence of a more complete technology stack for industrial AI.
This architecture connects several technology layers:
- AI infrastructure: GPUs, computing systems, networking, storage, power, and cooling.
- Foundation models: Models that provide general-purpose intelligence and support specialized industrial applications.
- Simulation and digital twins: Virtual environments for engineering, testing, and operational analysis.
- AI runtime platforms: Software environments that support the deployment and execution of AI applications.
- Edge computing: Local processing capabilities that bring AI functions closer to machines and production equipment.
- Physical AI: Robotics, autonomous machines, and intelligent systems that interact with the physical world.
- Safety and governance: Validation, access control, monitoring, and operational constraints for AI-enabled systems.
These layers must work together if industrial AI is to move beyond isolated pilot projects.
For example, an intelligent robotic application may require a simulation environment for training, a computing platform for model development, edge hardware for local execution, industrial communication for equipment coordination, and independent safety functions for machine protection.
A weakness in any one of these areas can limit the overall application.
This is why industrial AI should be approached as a systems engineering challenge rather than simply a software implementation project.
What Industrial Companies Should Prepare for Next
For manufacturers, system integrators, and industrial automation suppliers, the transition toward AI-native infrastructure creates several practical priorities.
First, existing automation architectures need to be assessed for their ability to exchange data with AI platforms. Legacy PLCs, DCS installations, SCADA systems, industrial networks, and supervisory applications may require additional integration layers before they can support new AI workloads.
Second, data quality and equipment information management deserve greater attention. AI applications depend on accurate process information, consistent equipment identification, historical operating records, and properly maintained engineering documentation.
Third, companies should establish clear boundaries between AI recommendations and actual control actions. An AI system may identify abnormal operating conditions or recommend changes to a production process, but direct interaction with control systems requires defined authorization, validation, and risk assessment.
Finally, industrial organizations should evaluate the long-term maintenance requirements of AI infrastructure. Computing hardware, cooling equipment, power distribution, network components, and software platforms all introduce lifecycle considerations that must be incorporated into maintenance strategies.
Conclusion: The Next Industrial Transformation Will Be Built on Integration
NVIDIA GTC Berlin 2026 demonstrated that industrial AI is moving toward a connected ecosystem spanning computing infrastructure, engineering software, simulation, robotics, and physical operations.
The announcements involving NVIDIA and industrial partners such as Schneider Electric illustrate how AI infrastructure is increasingly intersecting with established industrial engineering disciplines.
However, the real transformation will not happen simply because manufacturers install more GPUs or deploy more AI models. It will depend on how effectively companies integrate AI with existing control architectures, industrial data, equipment, engineering processes, and safety requirements.
For automation engineers, this creates both an opportunity and a responsibility. The next generation of industrial systems will require professionals who understand not only AI technologies, but also the machines, control strategies, communication interfaces, and operational constraints that determine whether those technologies can work in real production environments.
The companies that approach AI through this integrated engineering perspective will have a stronger foundation for moving from experimentation to sustainable industrial deployment.
