Qualcomm’s 70% Rally May Be Just the Start

While Qualcomm (QCOM) stock has largely been left behind in the AI-driven semiconductor rally over the last few years, the tide appears to have turned for the company, with the stock up by close to 70% over the past month as Qualcomm is positioning for the next phase of AI, one that moves beyond centralized compute and into billions of connected devices. In our previous analysis, we outlined how Qualcomm stock could surge 2x led by edge AI. Here, we outline the specifics of the shift and how Qualcomm’s customers mix changes.

Image by Cristian Ibarra from Pixabay

Why Is Qualcomm uniquely positioned for inference?

Today’s AI ecosystem runs on centralized compute, with workloads concentrated in data centers routed through the cloud and processed at scale by chips from companies like Nvidia (NVDA). That model has a ceiling. Routing every inference through the cloud is costly, latency-heavy, and power-intensive, and it simply does not scale to billions of devices. The next phase is local inference: AI running directly on the device, faster, more private, and independent of connectivity. Model compression techniques, including quantization, pruning, and distillation, are accelerating this shift by shrinking models without meaningfully sacrificing performance. The irony is that every efficiency breakthrough in AI research simultaneously makes Qualcomm’s edge hardware more capable.

This plays to Qualcomm’s strength.

For decades, the company has engineered for the two constraints that define edge AI: power efficiency and connectivity. Its Snapdragon and Dragonwing platforms distribute AI workloads across three specialized engines. The NPU handles the high-volume matrix math of AI at up to 80 TOPS (a measure of how many AI calculations the chip can run per second), at a fraction of the power draw of a laptop CPU. The GPU manages visual processing for generative AI. The CPU handles application logic. Each engine does only what it is built for.

Qualcomm also integrates the modem directly into the same silicon as the compute engines, so the device handles connectivity and inference on a shared power budget rather than running two separate chips. That integration represents years of co-design that a new entrant cannot replicate. Finally, edge devices have no HBM, the high-bandwidth memory that data center chips rely on to feed models at speed. Qualcomm’s platforms are built around that constraint from the ground up, while most competitors are trying to adapt data center thinking to the edge.

Who Are The Customers?

The “customer” for Qualcomm is shifting from just a few smartphone OEMs to an entire ecosystem of industrial and automotive giants.

Handset vendors remain the biggest customers for Qualcomm, accounting for 66.4% of chip revenue over the most recent quarter, but the revenue share has been shrinking. On the other hand, automotive accounted for about 14.6%, and the Internet of Things – which includes PCs and other chips – accounted for about 19%, with both segments growing. See Qualcomm’s key metrics

PC OEMs like Dell, Lenovo, and HP are choosing Qualcomm’s Snapdragon X Elite for their flagship AI PCs because it is the only NPU that meets Microsoft’s (MSFT) strict Copilot+ performance standards while delivering 20+ hours of battery life. Microsoft effectively set the qualification bar, and Qualcomm is the chip that clears it. That is not a feature win; it is a platform lock-in enforced by someone else’s standard.

In automotive, Volkswagen, BMW, and GM are anchoring their next-generation fleets on Qualcomm’s digital chassis. The reason is architectural. A car identifying a pedestrian cannot wait for a round trip to the cloud. Inference has to happen on the vehicle, in milliseconds, at all times. Qualcomm’s automotive design-win pipeline stands at $45 billion, built on contracts with lead times measured in years. Auto design cycles run five to seven years. Every win today is revenue that competitors cannot touch until the next vehicle generation. Qualcomm’s automotive business alone represents a $45 billion design-win pipeline.

On the factory floor, Qualcomm’s Dragonwing platform is powering robots that navigate and perceive their environment without a live server connection. The industrial edge is where physical AI becomes real, and Qualcomm is positioning itself as the default silicon for that transition, from Arduino prototypes to full-scale autonomous manufacturing systems.

Opportunities like Qualcomm highlight how individual stocks can rerate sharply, but they also come with concentrated risk tied to cycles and execution. A disciplined portfolio approach helps smooth these risks while still participating in long-term growth themes. The Trefis High Quality (HQ) Portfolio has consistently outperformed its market benchmark since inception, delivering total returns of 105%.