Qualcomm had long been identified with smartphone processors, but its AI data-center push gathered pace in June when it unveiled the Dragonfly C1000 server CPU. The company positioned the chip for agentic AI, arguing that strong computing performance need not come at the cost of soaring power use.¹

That backdrop matters because AI infrastructure is no longer solely a GPU story. Nvidia’s chips still dominate the demanding work of training and running major models, yet CPUs are increasingly important for the broader systems surrounding those workloads. Nvidia’s own AI infrastructure chief, Dion Harris, said in March that “CPUs are becoming the bottleneck in terms of growing out this AI and agentic workflow.”¹

Against that opening, Qualcomm and Amazon announced a collaboration across “multiple generations of customized silicon” for AWS AI infrastructure, with a particular focus on inference. The companies also plan to develop “high-performance optical connectivity solutions,” while Qualcomm will use Amazon AI servers to accelerate its own chip-design work.²

The financial structure underscores both the scale of Amazon’s potential commitment and the uncertainty still embedded in it. Qualcomm issued AWS warrants for 25 million shares at $161.26 each—up to $4 billion—vesting in stages tied to commercial arrangements and purchases of as much as $60 billion in Qualcomm server chips and other technology.²

Investors initially treated the pact as a meaningful breakthrough: Qualcomm shares rose 4% after the announcement. But it is better read as a foothold than a victory. AWS gives Qualcomm a hyperscaler endorsement and a path to its stated $15 billion data-center-sales target for fiscal 2029; converting that path into sustained orders is what would make it a genuine challenge to Nvidia.²