Robotics is experiencing an unprecedented era of innovation, driven by the rapid rise of Physical AI and machine learning capabilities. As robots increasingly break out of controlled factory cages and navigate dynamic human environments, traditional development paradigms are shifting. According to the newly released Inside the Robot: Architecture Benchmark Report by QNX, which surveyed 1,000 global robotics developers, hardware is no longer the primary constraint. Today, software has taken the lead: almost one in three developers (27%) cite software architecture and integration as their biggest performance bottleneck, compared to just 16% who cite hardware constraints.
Here is a breakdown of the core findings and what they mean for the future of robotics engineering.
The Physical AI Imperative vs. Real-World Readiness
The ambition across the robotics sector is clear. Nearly nine in ten developers (89%) view Physical AI, systems that can perceive, reason, and act autonomously in the physical world, as critical to their organizational strategy over the next three to five years. To support this transition, 85% of developers expect the role of software to expand, with development teams anticipating their largest future budget allocations will go toward AI-driven decision making (51%) and cybersecurity (51%).
However, ambition is currently outpacing field readiness. While industry optimism remains highly resilient, only 29% of developers feel "very confident" in their systems' ability to make safe, predictable decisions in unconstrained, real-world environments.
The Determinism Disconnect: Flexibility vs. Safety
As robotics systems increasingly operate alongside humans, a reality for 83% of respondents today, the demand for flawless reliability skyrockets. An overwhelming 95% of developers state that deterministic, real-time execution is essential to the systems they build.
Yet, the survey data reveals a startling architectural disconnect. Despite these stringent safety and execution demands, 91% of development teams still rely on general-purpose operating systems (GPOS), such as Linux, to run at least some real-time or safety-critical workloads. Nearly a third (32%) do so extensively.
To manage this, 64% of businesses currently utilize hybrid architectures that combine safety controllers, vision computers, and microcontrollers. While these hybrid setups enhance capability, they significantly increase integration complexity, which 42% of developers cite as their biggest software development challenge. Relying on non-deterministic system behavior introduces severe downstream consequences: developers report increased safety risks (46%), system instability (44%), and an expanded testing burden (35%).
The Certification Wall
The tension between utilizing flexible, open-source software and meeting strict industry safety guarantees is resulting in massive deployment bottlenecks. Two-thirds (66%) of developers report experiencing project delays specifically due to industry certification requirements. This friction is heavily tied to two main regulatory hurdles: cybersecurity standards (such as ISO/SAE 21434) and functional safety standards (like ISO 10218), which are cited as the most challenging to navigate by 51% and 49% of respondents, respectively.
The commercial and financial pressure to push through these bottlenecks is immense. Alarmingly, 84% of developers agree that strict deadlines and budget pressures can lead engineering teams to compromise on critical aspects of development, including safety protocols.
Implications: Re-evaluating the Foundation
The robotics industry is rapidly approaching an architectural tipping point. Scaling Physical AI requires foundational software that can handle mixed-criticality workloads safely, without crippling development velocity or risking failed compliance audits.
Recognizing the limits of their current stacks, 86% of respondents currently using a GPOS express openness to changing their operating system. When looking ahead to mitigate integration and scalability limits, many developers are shifting toward more robust paradigms: 31% cite a safety-certified commercial off-the-shelf (COTS) operating system as the best fit for their needs, while 27% are opting for virtualized Linux combined with a Real-Time OS (RTOS).
Conclusion
Hardware advances have set the stage, but the next major leap in robotics will be dictated by software architecture. Re-evaluating foundational architectures from the OS up is no longer just a backend technical exercise; it is a strategic necessity for securing certifications, ensuring human safety, and successfully scaling Physical AI into the real world.