Table of Contents
Author’s note: Qualcomm provided travel, lodging and other accommodations associated with Dragonwing IoT Day.
Through strategic acquisitions and a developer-first approach, Qualcomm is positioning Dragonwing to turn common edge AI capabilities into industry-specific solutions
SAN DIEGO —In his keynote at Dragonwing IoT Day, Qualcomm EVP and Group GM for Automotive, Industrial and Embedded IoT, and Robotics Nakul Duggal juxtaposed two companies with fundamentally different businesses, Zebra Technologies and Caterpillar. The former develops technologies for frontline workers and enterprise operations; the latter manufactures heavy equipment used in construction, mining and other industrial environments. Yet both increasingly depend on similar capabilities, including real-time situational awareness, local data processing and making intelligent decisions based on what is happening in the physical world.
That commonality is central to Qualcomm’s strategy for its Dragonwing platform. Rather than building entirely separate technology stacks for individual industries, Qualcomm is taking a horizontal approach spanning connectivity, heterogeneous compute, sensing, AI and software. These foundational capabilities can then be configured to address the distinct operational requirements of individual vertical markets.
Duggal acknowledged the central challenge that has long encumbered technology vendors trying to scale IoT offerings across a broad range of highly-verticalized end markets. “We’re a chip company,” he said. “A lot of this work that we’re talking about ultimately has to be deployed on chips, and chips are complicated because they’re all designed for many different things. The complexity of IoT is really, ultimately, being able to find a good product market fit.”
He continued: “As the world starts to think about tech in the context of industrial transformation/digital transformation, you start to say, ‘I’m going to need chips I can rely on.’”
The common architecture beneath industrial fragmentation
Historically, industrial IoT was largely concerned with connecting previously disconnected equipment and moving telemetry into IT systems. Connected machines generated operational data, but much of the intelligence required to interpret that data and determine an appropriate response resided elsewhere.
The introduction of sophisticated edge computing solutions changes that relationship. Across factories, mines, retail stores and warehouses, Qualcomm sees a recurring architecture in which sensors, cameras, robots and drones generate information that can be processed by intelligent endpoints, AI gateways, on-prem appliances and cloud infrastructure. The goal is to translate physical data into real-time operational decisions and, increasingly, autonomous actions.
While the applications and industries vary, the underlying requirements for connectivity, perception, inference, low-latency execution, security and reliability are substantially similar. This common foundation does not imply a uniform hardware configuration. A connected industrial sensor has fundamentally different processing requirements from a machine-vision system or autonomous mobile robot. Dragonwing consequently encompasses a broad silicon portfolio spanning microcontrollers, application processors, AI accelerators and connectivity technologies.
Qualcomm is also extending capabilities developed for smartphones and automotive systems into industrial environments. These include high-performance, power-efficient processing, image signal processing, sensor fusion and heterogeneous computing. Industrial applications introduce additional requirements around operating temperatures, functional safety, supply-chain resilience and extended product lifecycles. The objective is to make these capabilities available as building blocks rather than requiring developers to assemble specialized technology stacks from scratch.
An acquisition strategy spanning the development to deployment lifecycle
Qualcomm’s horizontal platform strategy has been substantially expanded through acquisitions aimed at addressing different parts of the intelligent device development and commercialization process.
The sequence began in March 2024, when Qualcomm acquired Foundries.io, a provider of embedded Linux development, deployment and lifecycle management technologies. Foundries.io brought capabilities for maintaining secure, updatable software on industrial devices over extended operational lifetimes.
In March 2025, Qualcomm announced its acquisition of Edge Impulse, adding tools for collecting and preparing data, training and optimizing edge AI models, and deploying and monitoring those models across different hardware platforms. The acquisition addressed the software development process that connects industrial data to usable intelligence.
On October 7, 2025, Qualcomm announced an agreement to acquire Arduino, gaining access to its community of more than 33 million developers and its widely adopted open-source hardware and software ecosystem. Arduino confirmed it had joined Qualcomm on October 31. The acquisition provides an accessible entry point for developers to prototype edge AI applications and subsequently migrate them toward commercially deployable systems.
Qualcomm subsequently expanded its vision portfolio through acquisitions of Focus.AI and Augentix. FocusAI added video intelligence and analytics capabilities, while Augentix contributed low-power imaging processors and associated expertise for intelligent cameras. Qualcomm included both companies among five strategic acquisitions in its January 5, 2026 industrial IoT portfolio announcement.
Qualcomm extended the software strategy beyond industrial IoT with the acquisition of Modular, announced on June 24, 2026, and completed on July 28. Modular’s AI software platform is designed to support efficient execution across heterogeneous compute architectures, including CPUs, GPUs, NPUs and custom silicon. Its relevance extends across Qualcomm’s edge and data center businesses.
Most recently, on September 23, 2026, Qualcomm announced an agreement to acquire PickNik Robotics, steward of the open-source MoveIt manipulation framework. PickNik extends Qualcomm’s robotics software capabilities from perception and navigation toward motion planning, manipulation and physical execution. Taken together, the acquisitions illustrate Qualcomm’s effort to address the full development lifecycle from prototyping, AI model development, optimization, deployment, operational management and, ultimately, physical-world action.
From horizontal technology to commercial scale
The acquisitions also support a significant change in Qualcomm’s commercial model. Duggal presented a comparison of Snapdragon, Dragonwing and Arduino that distinguished the company’s established high-volume, high-touch OEM relationships from the high-mix, lower-volume requirements of industrial IoT. Supporting thousands of specialized industrial applications requires a different engagement model from developing platforms for a relatively concentrated set of major smartphone manufacturers.
Qualcomm’s answer is to simplify access to silicon and software while enabling its developer and partner ecosystem to handle increasingly specialized implementation requirements. The company’s Qualcomm Linux 2.0 platform provides a common software foundation across Dragonwing processors, with modular capabilities for different applications. The developer ecosystem includes evaluation kits, AI development tools, software development kits, sample applications and agentic development tools. At Dragonwing IoT Day, Qualcomm described an ecosystem of more than 200 partners and 35 distribution partners supporting the transition from prototyping to commercial production.
Manny Singh, vice president of product management at Qualcomm, summarized the objective: “The overall idea is continuity.” That continuity must extend beyond initial development. Industrial customers need confidence that products can be manufactured, secured, maintained and updated over operational lifetimes that may span a decade or longer. “If there is one common theme across all of this, it’s there [is] still a lot of disconnected pieces…There is so much to navigate. So the question is, how do you make this thing simple?”
Customer examples illustrate how the common foundation translates into different operational outcomes. Cognex introduced its Dragonwing-powered In-Sight 3900 machine-vision system to support high-speed industrial inspection. At the event, Cognex CEO Matt Moschner described inspecting 2,000 coffee cans per minute, a workload that leaves approximately 30 milliseconds for each inspection. The requirements for latency, accuracy, thermal management and ruggedized deployment make local AI processing essential.
Zebra Technologies presented a different application of the same underlying capabilities. Its on-device AI software supports frontline workflows, including automating proof-of-delivery verification. The potential outcome is one or two additional deliveries per driver per day, illustrating how embedded intelligence can translate directly into operational productivity. The applications are distinct, but the architectural requirements are comparable.
For Qualcomm, the commercial opportunity depends on making those commonalities repeatable. At its June 2026 Investor Day, the company established a fiscal 2029 revenue target of $8 billion for industrial, networking and robotics, part of its broader diversification strategy.
Achieving that ambition will require demonstrating that Qualcomm technology can operate across different industries and that a common foundation can reduce development complexity, accelerate deployment and support commercially sustainable products despite considerable variability in end-market requirements. Industrial IoT will remain fragmented because its applications are inherently specialized. Qualcomm’s strategy is to standardize the technological foundation beneath that specialization.
