Edge AI Is Becoming a Standard Industrial Computing Requirement
Neousys Technology is seeing a clear shift in how industrial customers evaluate computing platforms. Edge AI is no longer limited to autonomous vehicles, defense systems, or other high-performance applications. Increasingly, manufacturers want AI inference capabilities directly inside compact industrial computers that previously handled data acquisition, control, or human-machine interfaces.
Neousys expects its edge AI-related revenue to increase by 10 to 20 percent year-on-year in 2026. Edge AI computers already represent approximately 40 percent of the company's product mix, covering fanless industrial computers, GPU-based systems, and rugged mission computers.
From an industrial automation perspective, this change is significant because it moves AI closer to the machine level. Instead of continuously transferring large amounts of raw data to centralized servers, industrial systems can increasingly process selected information locally and respond within much shorter timeframes.
Factory Safety Is a Practical Entry Point for Industrial AI
One of the most convincing applications is factory safety monitoring. Neousys and its partners recently demonstrated an AI-based system using NVIDIA Jetson technology to identify workers entering hazardous areas and generate real-time warnings.
This application illustrates where edge AI can provide a practical advantage. A safety system cannot always depend on cloud connectivity or remote processing. Local inference allows cameras and computing hardware to detect events near the source, reducing communication requirements and potentially improving response time.
However, AI should not automatically replace conventional safety systems. In applications involving personnel protection, AI-based detection should normally complement established safety circuits, interlocks, emergency-stop functions, and certified safety control architectures rather than becoming their sole protection layer.
Semiconductor Manufacturing Is Creating Strong Demand
Semiconductor manufacturing is another important growth area. Neousys stated that products used in wafer fabrication equipment generate approximately 10 percent of its total revenue, with the company supplying the world's two largest foundries.
This market requires compact computing platforms capable of handling machine vision, inspection, data processing, and equipment control. AI inference can become particularly valuable when inspection systems need to analyze increasingly complex visual information without transferring every image to centralized infrastructure.
The broader trend is toward distributed intelligence inside production equipment. Industrial PCs are therefore evolving from conventional control and data collection platforms into computing nodes capable of supporting vision algorithms, AI models, and real-time equipment analytics.
AI Infrastructure Is Expanding the Role of Industrial Computers
The expansion of AI data centers is creating an unexpected application for compact industrial computers. Neousys POC-700 series systems are reportedly being used by server manufacturers as controllers for coolant distribution units.
These systems monitor parameters including temperature, water pressure, and flow in in-rack and in-row cooling infrastructure. This application demonstrates that industrial computing is not limited to factories.
As AI servers generate higher thermal loads, cooling infrastructure becomes increasingly dependent on continuous monitoring and control. Compact industrial computers can provide localized data acquisition, communication, and supervisory functions within these distributed systems.
Autonomous Vehicles Are Connecting Automation With Edge Intelligence
Neousys is also expanding into autonomous mobile robots, automated guided vehicles, and other autonomous vehicle applications. Projects in Asia include outdoor delivery applications, while additional applications involve agriculture, law enforcement, and suspect tracking.
The industrial automation connection is straightforward. Autonomous machines require local processing for perception, navigation, sensor fusion, communication, and decision-making. Moving these functions closer to the vehicle reduces dependence on centralized computing infrastructure.
Regulatory developments could further accelerate this market. As countries establish frameworks for autonomous delivery vehicles, demand for industrial-grade computing platforms capable of operating outside controlled factory environments may increase.
Defense Is Becoming a Major Growth Driver
Defense applications currently account for approximately 20 percent of Neousys' revenue and are expected to reach about 25 percent by the end of 2026.
The company supplies computing platforms to drone manufacturers and system integrators developing uncrewed ground vehicles. These applications typically require ruggedized computing, local processing, extensive I/O capability, and operation under demanding environmental conditions.
From an engineering standpoint, defense applications also reinforce the importance of hardware longevity and platform stability. Industrial computers deployed in field systems often require longer product availability and predictable hardware configurations compared with conventional commercial computing platforms.
Intel's Next-Generation Platforms Could Push AI Into Smaller Systems
Neousys is preparing products based on Intel's next-generation Nova Lake platform, with an initial launch potentially arriving in the first quarter of 2027.
The important technical development is the platform's integrated neural processing capability. Dedicated NPU resources can allow smaller, task-specific AI models to execute locally without requiring the same level of GPU hardware traditionally associated with AI workloads.
This could have a meaningful effect on industrial automation. Many industrial applications do not require large generative AI models. They need relatively compact inference models for object detection, anomaly recognition, classification, predictive inspection, or equipment monitoring.
Lower computing requirements could therefore make AI practical in a much broader range of PLC-adjacent and IPC-based systems.
Healthcare Represents the Next Expansion Opportunity
Neousys plans to enter healthcare applications in 2027, focusing on AI medical imaging, patient monitoring, and care-related applications.
The company intends to apply image processing, AI computing, and rugged edge computing technologies to these markets. Healthcare also presents a different commercial environment because equipment typically has longer product cycles and can be less sensitive to short-term economic fluctuations.
For industrial computer manufacturers, this diversification is strategically important. The same hardware technologies used for machine vision and industrial monitoring can potentially support other environments where continuous local processing and long-term hardware availability are required.
Industrial Automation Is Moving Toward Distributed Intelligence
The most important takeaway from Neousys' expansion is not simply the growth of AI-enabled industrial PCs. The deeper change is the gradual redistribution of computing intelligence throughout industrial systems.
Traditional automation architectures often separated control, visualization, data collection, and higher-level analytics into relatively distinct layers. Edge AI is weakening that separation. A single industrial computer may now perform image processing, AI inference, protocol communication, data acquisition, and application-level decision support.
This does not mean that centralized systems will disappear. Instead, industrial architectures are likely to become more distributed, with each computing layer handling the workload best suited to its processing capability, latency requirements, and reliability constraints.
Engineering Perspective: AI Adoption Must Follow the Application
The growth figures are impressive, but industrial AI adoption should not be measured simply by the number of AI-enabled computers installed.
The more useful engineering question is whether local AI solves a specific operational problem. Applications such as worker detection, optical inspection, autonomous navigation, and equipment monitoring have clear reasons for using edge inference.
By contrast, adding AI to an industrial controller without a defined workload can increase system complexity without producing measurable operational benefits. Engineers should therefore evaluate processing requirements, latency, network architecture, model lifecycle, thermal design, cybersecurity, maintenance procedures, and hardware availability before selecting an edge AI platform.
The direction is nevertheless clear. Industrial computers are evolving from passive data-processing equipment into distributed computing platforms that can interpret information locally and support faster machine-level decisions.
Market Outlook
Neousys reported first-half 2026 revenue of approximately NT$1.14 billion, representing 36 percent year-on-year growth. The company expects full-year revenue to increase approximately 40 percent from NT$1.64 billion in 2025 and is targeting double-digit growth in 2027.
Edge AI, defense, smart manufacturing, and US market demand are expected to remain important contributors.
For the wider industrial automation market, the significance extends beyond one manufacturer. As AI inference becomes more efficient and industrial computing platforms become more capable, AI functionality is likely to appear in progressively smaller machines and control systems.
The next phase of industrial automation will therefore not simply be about connecting more devices. It will increasingly involve giving individual machines enough local intelligence to interpret their environment and act on relevant information in real time.
