Sensor-enabled real-time monitoring increases visibility across environments, informs digital asset or process twins, and fuels advanced analytics and model-based simulations. The development of smart sensing integrates sensing, processing, and communication in a single device. With this, sensors can measure physical parameters (e.g., temperature, pressure) and process the data locally, permitting real-time data analysis, filtering, and decision-making at the sensor level. For process industries and hazardous environments, new developments in Ethernet-APL enable connectivity and digitization of field devices through 2-wire Ethernet. Mobile robots can also serve as monitorsacross multiple assets where low-frequency data is sufficient.

Cloud adoption is increasing in business-critical functions, and Gartner predicts more than 70% of enterprises will use industry cloud platforms by 2027, up from less than 15% in 2023.2 Because cloud computing is elastic with market offerings for compute on demand, enterprises can focus the workforce and capital on core engineering and manufacturing competencies, rather than investing in on-premise data centers and the talent needed to run and maintain them. Today, manufacturers are exploring cloud suitability for business-critical manufacturing applications, as well as the potential in hybrid infrastructure capabilities. Furthermore, organizations are recognizing the untapped value of industrial data stored on the cloud, in data historians, and other systems.

New chips and hardware capabilities enable advanced analytics and processing at the edge, and to access the value, many manufacturers are increasingly adapting established IT principles (e.g., DevSecOps) for the OT environment. The benefits of edge inference include low latency for real-time insights, as well as cost efficiency, as only critical data for model training or historical analysis is sent to the cloud. Manufacturers today can design solutions balancing cloud and edge compute to optimize system performance and cost.

Numerous types of AI and automation are maturing. Advanced analytics capabilities inform decisions across the manufacturing value chain. There may also be opportunities to automate some decision-making, helping to streamline processes and enhance flexibility and agility. For several years, leading manufacturers have investigated how to integrate predictive AI outputs into closed-loop control systems. While this approach is being adopted cautiously, it has the potential to enhance control systems, and SDM can facilitate integration, enabling near real-time adaptability based on the advanced insights. The advent of Generative AI gives enterprises new tools to explore and summarize information across different data sources (including knowledge captured from experienced employees) and generate text, imagery, video, code, and even 3D models based on that data. Finally, the emergence of agentic AI opens the aperture of what’s possible with automation in both digital and physical worlds.

Most incumbent PLC vendors have publicly committed to bringing open and virtualized PLCs to market. Engineers cn administer new functionality remotely and for a number of PLCs at once. Virtual PLCs have increased connectivity across more systems within the ISA-95 stack. Full software-enabled PLCs allow for more secure continuous integration continuous delivery (CI/CD) pipelines and DevSecOps to build applications and capabilities that are decoupled from the physical environment.

Broadly, because a cyber-physical system combines interaction or control of physical processes with computation, networking, and software, new security vulnerabilities may result from rising data volumes, the increasing breadth of users with access to data, and connected IT/OT data across layers that have historically been segregated. A shift away from the ISA-95 stack and toward event-centric system integration leverages a flexible and scalable approach where data is shared through events, making it easier to manage and secure.

Advancements in robotics include AI-enabled software-defined control, learning, and simulation, as well as self-optimization via closed feedback loops. Communication with robots has been simplified with the development of generalized interfaces (e.g., chatbots, low- or no-code programming), while the ability to move AI and advanced processing to the edge and in the chip enables low-latency adaptable workflows. Processes are changing to accommodate robotic production capabilities, such as roboforming, rapid liquid printing, and additive and subtractive manufacturing. Meanwhile, humanoid robotic technology is becoming commercially viable, as the cost of humanoid robots has significantly decreased and advancements in AI and hardware have expanded flexibility and adaptability for a variety of tasks.

Today’s connectivity and protocols change how manufacturers physically connect assets, how they make sense of the data, and how that is communicated across systems and the workforce. Connecting more manufacturing assets while maintaining network performance gives manufacturers the ability to transfer, analyze, and understand data in real time. Advancements in 5G, in particular, enhance connectivity with ultra-low latency and high reliability. Whether via wireless standards or wired connectivity, assets, sensors, and edge devices communicate over a range of protocols, such as OPC UA (Open Platform Communications Unified Architecture) and MQTT (Message Queuing Telemetry Transport). These protocols support interoperability, scalability, and secure communication, which are essential for integrating Internet of Things (IoT) devices, edge computing, and cloud services.

There are a number of areas where manufacturers can leverage simulation technology to proactively and systematically finetune designs and operations in a digital space, avoiding the cost, risk, and time of testing and iterating in a physical space. Scenario simulations test process and product life cycles; engineering simulations test how machines and materials will interact; and digital twins enable simulations to provide “what if” scenarios that allow for better testing in a virtual environment, including for products, processes, and plant layouts. One challenge is that while many enterprises already use digital design tools, more data is needed to enrich the model for use in simulations and MES-style instructions. The challenge is to ensure a product meets design requirements and that it can be manufactured in a reasonable time frame with commensurate effort. Generative AI can be a valuable complement to simulations, as users rely on natural language prompts and low-code environments to manipulate or adjust digital twins.