QNX deepens NVIDIA ties
- QNX announced deeper integration with NVIDIA to support safety-critical edge AI in robotics, medical, and industrial devices. - The collaboration targets real-time, device-side inference suitable for latency-sensitive workloads. - That reinforces device-side processing as feasible for meeting features like local diarisation and privacy-preserving pre-processing. (x.com)
Artificial intelligence at the edge means running models inside the machine, not in a distant cloud server. On April 20, QNX said it is expanding its work with NVIDIA so that kind of on-device AI can run in safety-critical robots, medical systems, and industrial equipment. (accessnewswire.com) The new piece is software integration: QNX OS for Safety 8.0 is being paired with NVIDIA IGX Thor and the NVIDIA Halos Safety Stack. QNX announced the move at Hannover Messe in Germany, a major industrial technology trade fair. (accessnewswire.com) QNX is a real-time operating system, the software layer that schedules tasks so brakes, motors, cameras, and alarms respond on time. QNX says its OS for Safety 8.0 is built for predictable timing, low latency, and certification in systems where missed deadlines can become safety failures. (qnx.software) NVIDIA IGX Thor is the hardware side of that stack: an industrial edge computing platform for robots, medical devices, and factory systems. NVIDIA says it is designed for high-performance AI, reliability, and functional safety in regulated environments. (nvidia.com) The pairing targets machines that need to think locally and act immediately, including autonomous mobile robots, humanoids, surgical robotics, medical imaging systems, and industrial automation platforms. QNX said the goal is to let developers combine real-time control and accelerated AI on one platform instead of stitching together separate systems. (accessnewswire.com) That approach fits workloads where sending data to the cloud adds delay or creates privacy problems. Device-side inference can keep audio, video, or sensor data on the machine for tasks such as speaker separation, local preprocessing, and immediate safety checks before anything leaves the device. (nvidia.com) QNX is coming into this push with a large installed base in regulated systems. The company said in December 2025 that its software was embedded in more than 275 million vehicles worldwide, a figure it cited as evidence of its track record in safety-certified software. (accessnewswire.com) QNX launched OS for Safety 8.0 in August 2025 and said it was designed to meet standards including ISO 26262 ASIL-D, IEC 61508 SIL3, and IEC 62304 Class C. Those are the rulebooks companies use when certifying software for cars, industrial controls, and medical devices. (accessnewswire.com) Investors treated the announcement as material to BlackBerry, QNX’s parent company. MarketWatch reported that BlackBerry shares rose more than 11% in Toronto trading on April 21 after the company disclosed the deeper NVIDIA integration. (marketwatch.com) The immediate test is whether equipment makers adopt the combined stack for products that have to pass audits, not just demos. QNX and NVIDIA are selling a way to keep AI close to the machine while preserving the timing and certification demands that cloud-first systems often struggle to meet. (qnx.software)
