Qualcomm pitches a new wave of data-center AI chips, naming Meta as a CPU customer

Qualcomm used an investor-focused event to outline its next generation of data-center AI chips, unveiling what it called the Dragonfly portfolio and positioning it for demand tied to artificial intelligence workloads. The company said it is targeting more than $15 billion in data-center revenue by fiscal 2029, reflecting its push to compete more directly in the servers and accelerators used to train and run AI models.

In the same presentation, Qualcomm named Meta as a customer for its CPU offerings in the data center. The report described Meta’s role as CPU customer, which indicates Qualcomm is looking beyond mobile and into the broader compute stacks that power AI systems. How Meta will integrate or evaluate the chips, and what specific platform configurations are involved, were not detailed in the information provided for this report.

The Dragonfly framing matters because data-center AI deployments typically require tight coordination between CPUs, GPUs or other accelerators, memory systems, and networking. Qualcomm’s approach, as presented here, suggests it wants to be part of the server compute layer rather than only supplying components that might sit at the edges of AI infrastructure.

For Meta, the CPU customer label is relevant to how it can scale compute for its AI initiatives. Meta runs large-scale AI across content recommendations, ads systems, and internal AI tooling, and it also supports model development and inference at substantial volume. While the report did not provide implementation details, using a specific CPU supplier can influence system design decisions and procurement timelines.

Qualcomm’s revenue target also points to the financial stakes of the strategy. Data-center products tend to carry longer sales cycles and higher expectations for performance, power efficiency, and supply reliability, particularly when customers plan platform refreshes across multiple hardware generations. By stating a fiscal 2029 data-center revenue goal of more than $15 billion, Qualcomm is making the case that Dragonfly can convert AI-related demand into sustained commercial scale.

There is, however, limited disclosure in the available account. The report does not specify the individual chip models in Dragonfly, performance targets, production timing, manufacturing process nodes, power envelopes, or the availability of reference systems to customers. It also does not clarify whether Meta’s engagement is tied to a specific chip release, a design win that may extend across multiple generations, or a broader evaluation relationship.

What to watch next is whether Qualcomm follows this outline with more concrete product information, including launch schedules and customer commitments that can be validated through additional corporate statements or procurement milestones. Investors and industry observers will likely look for further details on how Dragonfly fits into server architectures and how quickly it can move from announcements into shipments.

For Meta, follow-through will likely come in the form of system-level deployments that show how CPUs from Qualcomm are paired with the rest of the AI compute stack. Any later clarification on the scope of Meta’s customer role, such as whether it is limited to certain workloads or data-center regions, would help determine how meaningful the Dragonfly partnership could become over time.