AI Software Becomes Top Priority for Robotics Developers, Forcing PLC Evolution

Why it matters now: The industrial automation landscape is approaching an inflection point. As robotics systems become increasingly software-defined and AI-enabled, the programmable logic controller (PLC) — long the undisputed backbone of factory-floor control — faces its most significant architectural challenge in decades. A newly released global benchmark study from QNX, a division of BlackBerry Limited, exposes a critical tension: robotics developers are racing toward AI-driven architectures, yet the real-time, safety-critical control layer that PLCs have traditionally owned is being stretched across software stacks never designed for deterministic operation.

The Inside the Robot: Architecture Benchmark Report, based on a survey of 1,000 robotics developers worldwide, paints a picture of an industry in transition — one where software is simultaneously the greatest enabler and the most stubborn bottleneck.

Software: From Enabler to Bottleneck

The headline finding from the QNX study is stark: 89% of robotics developers say Physical AI — embodied intelligence that drives real-world robotic actions — is critical to their future plans. Yet the same cohort identifies software as the single biggest inhibitor to robotics innovation. Looking ahead, 85% of developers expect software to play an even greater role in robotics over the next three to five years.

This is not a marginal shift. It represents a fundamental reordering of priorities on the development floor. Where mechanical design, power systems, and kinematics once dominated engineering attention, the software stack — operating systems, middleware, AI inference engines, and cybersecurity layers — is now absorbing the bulk of strategic investment.

The GPOS Paradox: What It Means for PLC Architects

Perhaps the most revealing — and concerning — data point from the study concerns the operating system layer. The research found that 91% of respondents run real-time or safety-critical workloads, at least in part, on general-purpose operating systems (GPOS). This is despite the same developers rating safety-certified commercial real-time operating systems (RTOS) as the best technical fit for their requirements.

This disconnect has direct implications for PLC-based control architectures. PLCs have historically provided the hard real-time, deterministic execution that GPOS cannot guarantee. But as robotics workcells become more autonomous — making AI-driven decisions at the edge, processing vision data, and adapting to unstructured environments — the clean separation between "PLC handles control, robot controller handles motion" begins to blur.

Why Developers Choose GPOS Over Safety-Certified RTOS

The QNX study raises a question that should resonate across every control engineering department: if safety-certified RTOS solutions are recognized as the best fit, why does GPOS usage remain so pervasive? The answer appears to lie in a combination of development velocity, ecosystem familiarity, and perceived complexity barriers.

Linux-based platforms, in particular, offer vast libraries, open-source AI frameworks, and a deep talent pool. For development teams under pressure to ship AI-enabled features, the path of least resistance often leads through GPOS — even when it means accepting non-deterministic behavior in the control stack. This trade-off may be acceptable in laboratory or pilot environments but becomes hazardous as Physical AI systems move into production environments alongside human workers.

The PLC Opportunity in an AI-First Robotics Era

For PLC manufacturers and system integrators, the QNX findings are not a threat but a roadmap. The study confirms that the market is actively seeking better real-time control foundations — it simply has not yet adopted them at scale. This creates a window for PLC vendors that can bridge the gap between traditional deterministic control and the AI-native software stacks developers are building.

Several developments point to where the industry is heading. Leading PLC platforms are already incorporating edge AI processing, OPC UA over TSN for deterministic communication, and software-defined I/O that can be reconfigured dynamically. The next frontier — already visible in demonstrations from companies like NEXCOM with its NVIDIA IGX-powered robot controllers — is the integration of functional safety, AI inference, and real-time motion control on unified hardware platforms.

Cybersecurity: The Equal-Priority Investment

Notably, cybersecurity tied with AI-driven decision making at 51% as the top investment priority among robotics developers. This equal weighting is significant and reflects the growing recognition that AI-enabled, connected robots dramatically expand the attack surface. For PLC systems that increasingly communicate with these robots over industrial Ethernet protocols, the cybersecurity posture of the entire control network is only as strong as its weakest node.

The study's finding reinforces what industrial security frameworks like IEC 62443 have long emphasized: security must be designed into the architecture, not bolted on afterward. A robot running AI inference on a GPOS without a safety-certified security foundation is not merely a performance risk — it is a potential entry point for adversaries targeting the broader operational technology (OT) environment.

What This Means for Control System Strategy

For manufacturing enterprises and system integrators planning their next automation investments, the QNX study offers three actionable takeaways. First, the software architecture decisions made today will determine competitive agility for the next five to ten years — the 85% of developers who see software's role expanding are not engaging in speculation; they are responding to a structural shift already underway. Second, the GPOS/RTOS gap represents both a risk and an opportunity: organizations that insist on safety-certified, deterministic foundations for real-time workloads will avoid costly retrofits as Physical AI systems move from pilot to production. Third, cybersecurity cannot be a secondary concern when robots are both AI-enabled and network-connected.

The PLC — far from being rendered obsolete by these trends — stands to become more strategically important than ever. But only if it evolves from a pure deterministic controller into a platform that can orchestrate AI workloads, enforce cybersecurity policies, and maintain hard real-time guarantees simultaneously. That is the challenge the QNX data lays bare, and it is one the automation industry must meet head-on.