On September 22, Qualcomm CEO Cristiano Amon stated that smartphones are moving away from apps and toward agents. This represents a shift in usage where AI agents understand user intent, remember context, and execute tasks across applications.
Instead of opening a maps app to check traffic, a messaging app to alert a friend they are running late, and a calendar to push back a meeting, a user simply tells their phone what they want, and the agent handles all three tasks on their behalf.
For this to occur seamlessly, a significant portion of the processing must happen on the device itself. Qualcomm chips power roughly half of the premium Android phones sold this year, according to Counterpoint Research, along with a growing number of Windows laptops, smart glasses, and cars. Any application of agentic AI within those devices will likely rely on Qualcomm hardware.
The company’s data center segment has been just as busy. In June, Meta signed an agreement for Qualcomm to supply CPUs for its next-generation server fleet. In September, Amazon signed a deal for Qualcomm to build custom AI inference chips for AWS.
Despite these developments, Qualcomm’s stock sits at roughly $173 at the time of writing, almost exactly where it was a year ago.
Krypton Research, which first wrote about Qualcomm in a post titled The Next Semiconductor Giant , believes the market still views Qualcomm merely as “the company in your phone.”
Their view is that AI inference is running into a memory wall and that Qualcomm has designed an innovative workaround that the market has not yet priced in.
Building on this point, looking at Qualcomm through a singular lens, whether its data center business or its phone hardware segment, overlooks why agentic AI is so vital to the company.
An agent lives in two places at once. It interacts with the user on a device while doing most of its heavy lifting in a data center. Qualcomm is one of the few companies supplying chips for both environments.
In this post, SixSigmaCapital and Krypton Research take a closer look at how Qualcomm fits into the agentic AI buildout, the victories it has already secured, and what still needs to go right.
Layout
Overview
Qualcomm in the Agentic AI Era
Getting Around the Memory Wall
Custom Chips and Connectivity
Where the Agent Meets the User
The Software Layer
Risks
Thesis
Conclusion
This post is for informational purposes only and does not constitute financial advice. Please conduct your own due diligence before purchasing any equities or assets discussed herein.