Is BlackBerry likely to become a critical infrastructure company for physical AI?

BlackBerry's strongest asset today is not cybersecurity and certainly not smartphones. It's QNX.
QNX is a real-time operating system already deployed in hundreds of millions of vehicles and safety-critical systems. As robotics and physical AI grow, many experts believe the industry will need deterministic, safety-certified software layers beneath the AI models. QNX is positioning itself directly for that role. BlackBerry has spent much of 2026 showcasing QNX as infrastructure for robotics, autonomous systems, industrial AI, medical devices, and humanoid robots. It has also expanded collaboration with NVIDIA around safety-critical edge AI.

The bullish thesis is:
1. AI models make decisions.
2. QNX executes those decisions safely in the physical world.
3. Physical AI requires reliability and certification that consumer operating systems weren't designed to provide.
4. BlackBerry earns licensing and royalty revenue as deployments scale.

However, even if QNX becomes important, BlackBerry still faces challenges:
1. QNX revenue remains relatively small compared with mega-cap AI companies.
2. Robotics adoption is still in its early stages.
3. The company needs to convert technical relevance into meaningful revenue growth.
4. Large industrial players and alternative RTOS (Real-Time Operating System) vendors also compete in this space.

The positive development is that BlackBerry's turnaround appears to be gaining traction. Recent reports indicate that QNX revenue growth has accelerated, with the royalty backlog approaching $950 million. BlackBerry's management has stated that the turnaround is largely complete, and the company is investing specifically into robotics and physical AI opportunities.

I think that the most plausible bull case is not that BlackBerry dominates AI itself. It's that QNX becomes a widely adopted safety and control layer for physical AI systems, producing steady royalty growth and a higher valuation multiple over time.

Curious to hear what others think. Do you see QNX as a foundational player in physical AI, or is this still too early to tell?