Following its major acquisitions of Augentix, Arduino, Edge Impulse, Focus.AI, and Foundries.ioFive companiesSubsequently, Qualcomm showcased its complete puzzle in the Industrial and Embedded Internet of Things (IE-IoT) field at CES 2025.
Qualcomm announced the launch of its new Dragonwing Q series processors, emphasizing that by integrating the technological resources from the aforementioned acquisitions, it can now provide a one-stop solution for global enterprises and independent developers, from underlying hardware to upper-level software services, aiming directly at the top of the core edge computing and AI technologies.
Dragonwing Q-8750: A Monster in Device-Level AI Performance
As the flagship product in the series, the Dragonwing Q-8750 is described by Qualcomm as "the most powerful IoT processor to date." Its built-in AI engine boasts a computing power of up to 77 TOPS (supporting INT4/8/16 and FP16), enabling it to run large language models (LLMs) with up to 110 billion parameters directly on the device, completely eliminating reliance on the cloud.
In terms of hardware specifications, the Dragonwing Q-8750 supports up to 12 physical lenses and three 4800-megapixel image signal processors (ISPs), clearly designed for high-end drones, media centers, and multi-angle vision systems.
Dragonwing Q-7790: A General-Purpose AI Computing Processor
The Dragonwing Q-7790 targets a wider range of consumer and industrial applications, such as smart cameras, AI TVs, and video collaboration systems. While positioned as a more affordable option, it still boasts 24 TOPS of AI computing power and supports dual 4K60 display output and AV1 hardware decoding. For security, it incorporates Total Management Engine and Secure Boot technology to ensure data integrity.
The benefits of the five major acquisitions are becoming apparent, with the Arduino ecosystem now included.
Another major focus of this presentation was how Qualcomm is digesting its five major acquisitions over the past 18 months:
• Augentix:This Taiwanese imaging chip manufacturer has integrated its technology into Qualcomm's product line, enhancing the low-power processing capabilities of IP cameras and vision systems.
• Arduino & Edge Impulse & Foundries.io:By integrating the Arduino open-source community, Edge Impulse's machine learning tools, and Foundries.io's secure deployment platform, Qualcomm emphasizes significantly lowering the barrier to product development, enabling developers to quickly move from prototyping to commercial deployment.
In addition, Qualcomm has launched a video intelligence service called Qualcomm Insight Platform and a ground positioning service called Terrestrial Positioning Service. The former uses large language models to analyze video data, while the latter uses Wi-Fi and base station signals to provide accurate positioning in environments without GPS.
Analysis: Qualcomm is not just selling chips, but also "solutions".
In the past, Qualcomm was mostly perceived as a "modified mobile phone chip" in the IoT field, but this time, through the new Dragonwing brand and a series of acquisitions, it is clear that it is aiming to build an independent and complete ecosystem. In particular, incorporating Arduino into its portfolio is a very clever move, allowing Qualcomm to directly reach millions of makers and developers worldwide, binding user habits from the source.
When the Dragonwing Q-8750 is able to run large language models with 110 billion parameters on drones, and developers can develop using familiar tools like Arduino or Edge Impulse, it highlights how wide and deep Qualcomm's moat in the edge AI market can become.
