The significant performance enhancement of industrial robots and smart devices will fundamentally change the way we use artificial intelligence at the edge, as well as our understanding of cloud and data centers.
Image Source: Gorodenkoff / Shutterstock
Humanoid robots, smart devices, and autonomous driving are often considered lucrative business cases in the edge domain. However, edge artificial intelligence computing will liberate AI from centralized servers in data centers and the cloud, bringing it to manufacturing workshops, operating rooms, and municipal centers, where data can be processed in real-time closer to IoT devices, sensors, and intelligent systems. It also provides low latency and autonomous decision-making capabilities, making AI ubiquitous and enabling fully autonomous industrial facilities, thereby transforming business and public life. This is a topic of great interest to Chris Nardecchia, CIO of Rockwell Automation, a customer and partner of NVIDIA, at the forefront of edge AI computing. “The shift of AI from cloud-centric architectures to edge-based deployments not only represents an evolution of technology,” said Nardecchia, “it fundamentally redefines how AI capabilities integrate into every aspect of our industrial and personal environments.” Last month, Rockwell Automation showcased its advanced factory-level virtual control testing technology, Emulate3D, at NVIDIA’s GTC conference. This solution integrates NVIDIA’s Omniverse API, allowing manufacturers to validate automation systems through virtual factory acceptance testing before physical deployment.
For example, edge AI allows businesses to deploy AI applications on smart devices or robots in warehouses, performing compute-intensive inference and reasoning models close to the data source, rather than through public cloud or data centers. This significantly accelerates the speed of AI.
At the recent conference, NVIDIA continued its ambitious push into the edge domain, launching a series of advanced AI hardware, software platforms, and developer frameworks, including the Jetson Orin, Xavier, and Nano platforms, enhanced Blackwell Ultra AI chips for advanced edge applications, the Groot N1 AI robot model for autonomous machines, the IGX Orin industrial-grade edge AI platform for industrial and medical needs, and the Nvidia AI data platform for data analytics at the edge. In particular, NVIDIA’s core EGX enterprise edge AI platform provides real-time AI workload support for healthcare, manufacturing, and retail industries, while its Metropolis platform powers edge video analytics for smart cities.
The Jetson Nano super workstation launched at the NVIDIA GTC conference will provide powerful AI capabilities within reach for commercial users in remote offices or business centers. Its healthcare-focused Clara, autonomous driving-focused Drive, and 5G network-focused Aerial platforms will also provide real-time monitoring, predictive maintenance, and process optimization throughout the lifecycle of industrial assets, reducing downtime and improving on-site system performance. An analyst stated that NVIDIA’s series of platforms and products targeting Physical AI highlights the vendor’s expansion from a mere semiconductor manufacturer to a provider of hardware, platforms, tools, and frameworks. “The market has yet to fully grasp the profound implications of NVIDIA’s move,” said Chirag Dekate, chief AI analyst at Gartner Group. “This is indeed astonishing. The combination of EGX and Jetson, along with NVIDIA’s Cosmos platform (where you can develop environments similar to Physical AI, combining the essence of AI and digital twins), creates an environment conducive to accelerating intelligent training that can be deployed at the edge. It is at the edge that they are now transforming our robotics, intelligent robots, autonomous vehicles, and humanoid robots. They are opening up a new growth driver, just as they did with GPUs in data centers.”
1. Sequence of Events
The market first embraced generative AI for content creation, then shifted to agentic AI capable of reasoning and executing tasks. However, the core of the industrial AI revolution is Physical AI or robotics that supports AI, enabling fully autonomous industrial facilities, most of which are deployed at the edge.
For instance, Rockwell’s autonomous mobile robots (AMRs) have demonstrated measurable impacts in throughput, labor optimization, and time savings. Nardecchia stated that these robots serve as mobile edge computing platforms, processing sensor data locally while feeding aggregated insights back to the company’s FactoryTalk Edge Manager. He noted that NVIDIA’s edge computing platform will expand, enrich, and reveal industrial data, creating new sources of value through applications and analytics. When combined with Physical AI and agentic AI, this capability becomes particularly powerful, enabling truly autonomous systems that can perceive, decide, and act with minimal human intervention. He added that Rockwell’s strategic acquisitions of Otto Motors and Clearpath Robotics position the company favorably in production logistics automation.
2. Edge AI is Transforming Industrial Digitalization
According to IDC’s recent forecast for 27 enterprise industries, global spending on edge computing solutions is expected to approach $261 billion this year, with a projected compound annual growth rate of 13.8%, reaching $380 billion by 2028. Dave McCarthy, vice president of IDC’s cloud and edge services research, stated, “Edge computing is poised to redefine how enterprises leverage real-time data, and its future depends on tailored industry-specific solutions that can meet unique operational needs.” “We see service providers doubling down on investments to build low-latency networks, enhance AI-driven edge analytics capabilities, and establish partnerships to provide scalable and secure infrastructure. These efforts are crucial to realizing the full potential of edge computing, supporting everything from smarter manufacturing workshops to responsive healthcare systems, ultimately driving a new wave of innovation across verticals.” Therefore,CIOs are planning their next-generation AI architectures, using powerful platforms, tools, and frameworks to create robots and IoT devices with such autonomous decision-making capabilities.
“CIOs are certainly planning AI for edge workloads,” said Nate Melby, CIO of Dairyland Power Cooperative, who is looking to leverage these advancements to manage the grid during storms and enable systems to perform rapid analysis and decision-making in hazardous environments. The new business opportunities and profit outcomes brought by using edge AI devices in other physically restrictive environments are also highly anticipated. Melby stated, “By pushing AI to the edge, we can build resilience by reducing reliance on centralized architectures and optimize and locally process more sensitive or critical task data by balancing cloud resources with local devices, enabling easier scaling and resource flexibility. However, this will take some time to develop.”
3. Growing Interest in the Edge
Many top cloud and AI providers, including OpenAI, Google, and Amazon, as well as innovative AI startups, are targeting the edge market. For example, cloud provider Oracle recently added GPU-optimized configurations to its Oracle Roving Edge Device. “We see customer demand for edge computing AI,” said Dave Rosenberg, senior vice president of field and industry marketing at Oracle Cloud Infrastructure (OCI).
Amol Ajgaonkar, chief technology officer of product innovation at Insight Enterprises, added that many industries beyond manufacturing will leverage edge AI, but it is not without challenges. “If the edge is defined as anything not in the cloud—laptops, machines in manufacturing workshops, point-of-sale devices in retail—then healthcare, retail, and finance are major targets for edge AI,” he said. “A significant challenge for edge AI is determining which data is good or bad for the current task and setting up processes so that an AI agent or group of agents can manage without continuous human intervention,” Ajgaonkar said. “When creating predictive models, for example, managing the ongoing maintenance of factory workshop machines, certain deviations or incorrectly formatted data, if not filtered, can affect the model and distort the agent’s final actions. Clean input data is always critical for clean output, but this is a delicate balance.” Tom Richer, CEO of AI consulting firm Intelagen and former CIO, suggested that CIOs closely monitor NVIDIA’s progress, as they dominate AI infrastructure, data center transformation, and edge AI capabilities, all of which directly impact organizational innovation and competitiveness.
He stated, “The increasing adoption of edge computing to handle AI workloads, driven by the demand for low latency, bandwidth optimization, and enhanced security, requires strategic decisions between local deployments and service provider deployments, with hybrid approaches often proving to be the most effective way to balance control and scalability. This requires CIOs to develop clear strategies, invest in necessary infrastructure, and keep pace with evolving technologies.”
4. Gaining an Edge from the Edge
Implementing Physical AI requires robust edge computing capabilities that can process sensor data with minimal latency and execute complex algorithms. Nardecchia from Rockwell explained that edge computing achieves decentralization by moving data processing closer to the data source, resulting in faster response times, higher data transmission efficiency, and enhanced security—all critical factors for robotics and industrial automation applications. As platforms and technologies continue to mature, we can expect AI to become increasingly embedded in the physical systems of industrial environments.
For companies like Rockwell, this evolution represents an opportunity to integrate edge AI capabilities into their entire product portfolio. The business outcomes of well-managed edge computing are significant, including affordable data access, faster software deployment, future-proof analytics platforms, improved security posture, better scaling of digital transformation initiatives, and reduced total cost of ownership (TCO). The Edge AI Foundation states that CIOs and enterprises are looking to achieve automation and intelligent devices at the edge. “The core of edge AI is running AI workloads where data is generated, and the pull to the edge means lower costs, lower power consumption, and often greater impact, which can also mean enhanced privacy, low latency, flexibility, and settlement,” noted Pete Bernard, CEO of the nonprofit organization, with CIOs responsible for formulating information strategies. “You want to move computing as close to the data generation point as possible, avoiding cloud ingress and egress fees and operational expenditure (OpEx) costs, while having more control over your processing processes.”
As platforms and technologies continue to mature, we can expect AI to become increasingly embedded in the physical systems of industrial environments.
“The rise of foundational models is spreading to the edge through distilled and quantized transformer models and small foundational models,” said Paul Golding, vice president of edge AI at Analog Devices. “This shift requires dense, compute-intensive infrastructure to be deployed at the edge. At the same time, the demand for real-time processing, low latency, and privacy is pushing AI closer to the data source—what we often refer to as the sensor edge or physical edge. The ability of agentic AI to learn, adapt, and act autonomously in real-time will fundamentally change task orchestration across heterogeneous nodes, as we move from machine automation to machine autonomy, giving rise to new forms of distributed intelligence that enable critical task operations to run at the edge without relying on centralized cloud systems. The frontier of AI remains vast.”
Top Ten Takeaways:
- Shift from Cloud to Edge:Artificial intelligence is shifting from relying on cloud data centers for processing to processing at the source of data generation (e.g., devices, machines, local servers), known as “edge computing.”
- Benefits:Edge AI can achieve extremely low processing latency, real-time responses, and autonomous decision-making, while potentially reducing data transmission costs and enhancing data privacy and security.
- Key Applications:Including industrial robots, smart home devices, autonomous vehicles, smart manufacturing, telemedicine, smart retail, and smart city management.
- NVIDIA’s Dominance:NVIDIA is not only a chip manufacturer but also a significant provider of hardware (e.g., Jetson series, Blackwell), software platforms (e.g., EGX, Omniverse, Metropolis), and development tools in the edge AI space.
- Physical AI:This is the next stage of AI development, focusing on endowing robots and autonomous systems in the physical world with AI capabilities to perceive their environment and act autonomously.
- Industry Adoption:Industrial giants like Rockwell Automation are actively adopting edge AI technologies (in conjunction with NVIDIA platforms) to achieve factory automation and robotics control.
- Market Growth:The global edge computing market is vast and is expected to maintain strong growth momentum in the coming years (according to IDC data).
- Other Players:In addition to NVIDIA, major cloud service providers (e.g., Oracle, Google, Amazon) and AI startups are also heavily investing and positioning themselves in the edge AI market.
- Challenges:Deploying edge AI requires addressing data quality issues (filtering valid data, avoiding biases) and how to enable AI agents to work effectively without continuous human supervision.
- Future Trends:More powerful AI models (e.g., foundational models) will be optimized and deployed to the edge; the rise of “agentic AI” with autonomous learning and action capabilities will emerge; systems will evolve from automation to “machine autonomy,” forming distributed intelligent networks.
Insight: A manufacturing plant deploys edge AI cameras and sensors on its assembly line. Utilizing AI models running on local devices (such as NVIDIA Jetson), it detects product defects in real-time and monitors equipment status. The system instantly marks defective products and predicts maintenance needs, responding quickly, only sending critical summary data to the cloud. This significantly enhances quality control efficiency, reduces downtime due to equipment failures, and optimizes the entire production process, demonstrating the advantages of edge AI’s low latency, high efficiency, and local processing.