Qualcomm’s Software Acquisitions May Finally Overcome Edge Friction (2025)

Qualcomm has announced its intention to acquire Edge Impulse, whose software enables the training and deployment of AI and machine learning models to small edge-based devices including gateways and microcontrollers. My colleague Bill Curtis brought this to my attention; as someone with personal experience bringing a software-defined edge solution to market, I am very intrigued and hopeful about Qualcomm’s news.

The deal is a follow-up to last year’s acquisition of Foundries.io, which gives Qualcomm the ability to provide containerized deployment and orchestration for edge-based devices. After decades of companies trying — mostly without success — to attract developers and extend software-defined scale and flexibility to the edge of the network, Qualcomm’s strategy may now have a real chance.

Why Building Software For The Edge Has Been So Challenged

A couple of significant factors have hampered progress in this area. First, the edge is very hard to scale. To put it another way, building robust networked products outside the datacenter is very hard. The edge market is vast and spans a wide range of industries. What works on an oil rig will not work in a hospital or on a truck. Each of these industries has unique and specialized requirements, certifications and lifecycles. There are also more rigid demands on reliability, performance and serviceability. (Think about how many of the edge devices on an oil rig have to function without interruption in the rain.) And, if that wasn’t hard enough, the go-to-market models are also highly fragmented. This level of complexity means that edge solutions are expensive and sold in low volumes. Aside from all the implications for hardware vendors, this is the opposite of the business model for the typical horizontal software or cloud provider, which creates a disincentive for developers (and investors) to build scalable business plans.

The second problem is that incumbent vendors have fought software disruption for decades. Given the high demands of this space, solutions have tended to be highly proprietary and vertically aligned. These are also typically hardware-based solutions with long lifecycles. As the broader tech industry knows from past experience, it’s easier to sell and service proprietary, high-dollar hardware than low-cost, standards-based software. Additionally, edge purchases have traditionally been treated as a capital expense and depreciated, so consumption-based models can be hard to sell. Therefore, the biggest vendors that service the edge have done a masterful job of slowing innovation and forcing customers to stick with the older solutions. Mind you, it’s not for a lack of trying. Hundreds of startups and even massive companies like Exxon and GE have launched software-defined technology standards, but to no avail.

Qualcomm May Finally Have The Goods To Modernize Software At The Edge

I see three factors working in Qualcomm’s favor here.

1. Small ML inferencing has become a powerful use case. Over the past two years, we have seen an explosion in AI and ML models in the marketplace. And while generative AI has been a major benefactor, machine learning and vision processing are also seeing great benefits from techniques such as model distillation and low-cost fine-tuning. Inference models now have the capability to process at much greater scale, leading to more precise actions and greater volume, which in turn leads to higher-value use cases. But for developers to take advantage of these new models in production, technologies like those developed by Edge Impulse and Foundries.io are necessary to deploy and manage these models.
2. Standardization favors an IT-influenced hardware and software platform. The software-defined edge is different than what we see with traditional IT or cloud platforms. Notably, an edge platform needs to contemplate specialized and proprietary hardware devices. For example, the low-latency requirements for a factory-floor safety system do not have a suitable replacement in the IT world. Software-defined edge platforms provide standardization for developers, operators and enterprises, thus reducing the costs and complexity of different commodity and specialized hardware. That said, despite these differences a software-defined edge platform can leverage technologies from other IT platforms. For instance, Foundries.io support of the Yocto open source project enables a consistent and properly sized Linux distribution across the enterprise. And support for UEFI provides a layer of additional security that devs can write to, versus building their own. The addition of Edge Impulse to this platform extends the standardization of traditional control applications to include AI and ML, which will drive further automation at the edge.
3. Qualcomm has an ecosystem advantage. The ideas and technologies from these acquisitions and Qualcomm’s strategy are not necessarily new. But what makes this different is that Qualcomm has an extensive ecosystem in place for these innovations to add further value. Platform plays in the past have either been driven by startups that lacked the broad ecosystem required to get traction, or by vendors that could not play the role of a neutral broker (e.g., Siemens would not necessarily want to adopt GE’s platform). This places Qualcomm in a position to bring together many groups including technology players, hardware manufacturers and industry specialists to create and deploy new solutions.

So, after decades of challenges, Qualcomm may finally be able to leverage its semiconductor chops — along with its newer software capabilities — to disrupt markets that could really use it. More modern software approaches can speed up upgrade and maintenance cycles, as well as mitigate the need for proprietary hardware. Longer term, this could also serve as the underlying fabric that enables new AI innovations in robotics and automated transportation. The friction at the edge may finally be overcome.