Regular readers of Manufacturing Tomorrow are fully aware of the drivers and benefits of automation in reducing costs and advancing operational efficiencies in manufacturing. Some of you are starting to add physical AI to your automation projects for an even greater return on investment. What I’m hearing in my daily conversations is that there is an important element that’s too often left out of the automation upgrade discussion – and that’s the underlying network that can make or break a successful project.

Before talking about the network, I want to discuss physical AI and how it is starting to be deployed in a factory setting. This will help clarify the need for robust and secure network connectivity. Physical AI takes the concept of digital AI that we’re all familiar with and applies it to real-world physical systems like robots, automated vehicles and sensors. With the addition of physical AI capabilities, these systems don’t just do repetitive pre-programmed tasks, but they take in information from their environment, learn and adapt in real-time based on this information, and make autonomous decisions based on this data analysis to continuously improve performance, safety and efficiency. 

Some of the common applications already being deployed today include predictive maintenance that uses sensors to forecast failures; automated quality control with computer vision defect detection; process optimization that analyzes real-time data to identify and apply improvements, inventory optimization for accurate demand forecasting, and automated forklifts and autonomous mobile robots for safely and efficiently moving material around a factory floor or warehouse.

Each and every one of these applications require one important element to deliver the promised improvement – a reliable, high-performance network. If the network that is supporting any of these applications goes down, efficiency gains can become losses in a very short period of time. The need for the physical system to act and respond in real time also means that the network needs to act and respond in real time, with the lowest possible latency between a signal and a response. The applications also require maximum throughput, especially in the uplink when robots, automated vehicles and sensors send information back to the network for rapid analysis and response. In the case of an automated forklift or mobile robot, there is potentially an additional worker safety concern if the response is slow or a machine stops dead in its tracks.

Something that is too often overlooked is that manufacturers are not monoliths – a single group facing a similar set of problems – but rather you are as varied as the types of products you produce and the materials they’re made from, the production process and the degree of customization required, and the size of your operations. The automation system you implement and the underlying wireless network that supports it need to be flexible enough to adapt to your environment and be future proof for new requirements as they develop. As physical AI becomes the new norm for industrial automation, the amount and types of factory optimization that can be applied will increase exponentially, and you need an underlying network architecture in place that can adapt and grow with your automation strategy.

For the first generation of factory automation systems, Ethernet cables or Wi-Fi were the connectivity options available. As automation systems further developed and became mobile, the first generation of private 4G/5G networks were deployed to improve connectivity. As new physical AI automation systems come onto the market, and they are currently being developed at the speed of AI, the next generation of private 5G is needed to greatly improve network performance and significantly reduce latency.

What is a next-generation private 5G platform? I have already touched on the key characteristics required. Instead of a small-cell based architecture that was first developed by mobile operators for their consumer customers, next-generation architectures are designed to take 5G to its highest performance limits, to deliver on the ultra-low-latency promise of 5G, and at the same time be easy to deploy and manage in an industrial setting. XCOM RAN is one of first next-generation private 5G solutions available on the market, so I wouldn’t be surprised if you’re not yet familiar with these unique capabilities of our solution.

XCOM RAN utilizes a multipoint radio system, with radios that jointly process signals at the network edge, creating the XCOM RAN Supercell design for high-capacity and scalable wireless connectivity. This edge compute design also means that the data collection, analysis, and response needed to apply physical AI to industrial automation applications are all happening adjacent to the wireless stack processing, enabling low latency, low jitter response, and, more importantly, enabling optimized wireless-aided algorithms. With XCOM RAN we have increased capacity by 4-6x over current private 5G offerings for flawless high-performance connectivity in the densest automation environments. Key elements of the solution are software-defined and deployed on COTS hardware – reducing cost, eliminating forklift upgrades, and enabling innovation at the speed of software. This architecture reduces the need for site surveys and network design, for a private 5G solution that deploys quickly and is easy to manage. And unique to all other solutions available on the market, XCOM RAN can leverage Globalstar’s licensed Band n53 as a dedicated band for worry-free private 5G deployments.

The time plan for the network underlying your physical AI automation program is now. According to a Sept 2025 report by the World Economic Forum, “Physical AI is not the distant future. Intelligent robotics is already transforming manufacturing, and momentum will only increase.”  If this isn’t a discussion you’re already having with both your IT and OT teams, it should be.

Tamer Kadous is the General Manager of Globalstar’s Terrestrial Networks business unit, where he leads technology creation, market strategy, sales, and business development. In this role, he is responsible for advancing and monetizing the business unit’s core assets, including XCOM RAN technology and Globalstar’s globally licensed S-band spectrum for terrestrial use (3GPP n53).

Prior to Globalstar, Kadous led wireless engineering at XCOM-Labs, where he focused on developing advanced 5G technologies for demanding use cases such as extended reality and industrial automation. He built and led the engineering organization, set technology strategy, and guided products from concept through commercialization, resulting in multiple commercially deployed wireless solutions serving specialized, high-performance environments.