Qualcomm has expanded its industrial and IoT-focused Dragonwing line with two new processors – the Q-7790 and Q-8750 – designed to bring high-performance generative AI to the edge. The new chips target drones, smart cameras, industrial vision, AI TVs, collaboration systems and other on-device inference workloads where cloud round-trips are too slow or too expensive.
What the chips deliver
The Q-7790 provides 24 TOPS of on-device AI performance, suited to applications such as smart cameras and media hubs that need real-time inference without cloud dependencies. The Q-8750 steps up to 77 TOPS with INT4/8/16 and FP16 precision support, enabling on-device large language models with up to 11 billion parameters – an industry-leading capability for an edge SoC.
Why edge AI now
Edge inference is one of the fastest-growing sub-categories in semiconductors. Industry analysts at Counterpoint Research expect edge AI shipments to enter a commercial breakthrough year in 2026 as device makers race to embed generative capabilities directly in products. Competing edge announcements from NVIDIA, Intel and Synaptics have crowded the field at events like CES 2026 and Computex.
Strategic implications
Qualcomm''s Dragonwing strategy positions it as a counter to NVIDIA''s Jetson Thor in physical-AI and robotics design wins, and to Apple silicon in on-device consumer AI. The new chips ship with built-in cybersecurity primitives, support for confidential compute, and SDK integration with Qualcomm''s broader AI stack.
Availability
Reference designs are sampling now with development boards expected at Embedded World 2026 in Nuremberg. Commercial OEM products powered by the Q-7790 and Q-8750 are expected starting Q4 2026.
Reporting based on coverage from Qualcomm, Counterpoint Research and Semiconductor Engineering.