Qualcomm secures Meta as first Big Tech customer for data center chips

Qualcomm just landed the kind of customer that turns a side project into a real business. Meta has signed on as the first major Big Tech buyer of Qualcomm’s data center CPUs, a deal that marks the chipmaker’s most significant step yet into a market it has been circling, and occasionally retreating from, for over a decade.

The multi-generation strategic agreement, announced at Qualcomm’s Investor Day in New York on June 24, centers on the Qualcomm Dragonfly C1000 CPU. Production is slated to begin in the second half of 2028, with the chip designed specifically for agentic AI workloads and optimized for high performance per watt.

From smartphones to server racks

The numbers tell the story. Qualcomm raised its fiscal 2029 non-handset revenue target to $40 billion. Of that, over $15 billion is projected to come from data center AI alone.

Meta CEO Mark Zuckerberg framed the partnership around the company’s infrastructure buildout for what he called “personal superintelligence.” In practical terms it means Meta needs enormous amounts of compute that runs efficiently at scale. The Dragonfly C1000’s focus on performance per watt fits squarely into that requirement.

A relationship with a complicated history

Qualcomm and Meta have actually tried this before. Back in the 2010s, the two companies tested Arm-based CPUs for data center applications. Those efforts were paused due to cost constraints and various technical challenges.

What changed was Qualcomm’s 2021 acquisition of Nuvia, a startup founded by former Apple silicon designers. That deal brought in the kind of chip design talent that had produced Apple’s M-series processors. The Nuvia DNA has already shown up in Qualcomm’s laptop chips, and now it’s being channeled into server silicon.

Qualcomm has also been busy building out its data center portfolio through other channels. The company recently acquired AI software startup Modular for approximately $3.9 billion, adding software capabilities to complement its hardware roadmap. And it has struck agreements with Saudi AI company Humain and ByteDance for custom ASICs, suggesting a broader strategy that goes beyond just selling off-the-shelf processors.

What this means for investors and the competitive landscape

The $15 billion data center AI revenue target for fiscal 2029 is ambitious, particularly given that the Dragonfly C1000 won’t even enter production until the second half of 2028. That leaves a narrow window for Qualcomm to ramp manufacturing, prove reliability at scale, and convert pilot deployments into sustained purchasing relationships.

The Modular acquisition adds an interesting wrinkle. Software ecosystems are often what determine whether a chip platform succeeds or fails in the data center. Nvidia’s CUDA moat is the textbook example. By acquiring Modular, Qualcomm appears to be building its own software stack rather than relying solely on open-source tools or third-party support.

Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.