Qualcomm Unveils Dragonfly Data Center Roadmap: A 250-Core C1000 Server CPU and AI300 Accelerator, With Meta Signed as a CPU Customer
At its June 24 Investor Day, Qualcomm detailed a multi-generation data center chip line built for agentic AI inference, anchored by the Dragonfly C1000 CPU and AI300 accelerator, and named Meta as a server-CPU customer.
Overview
Qualcomm laid out a comprehensive plan to enter the AI data center market on June 24, 2026, unveiling a chip line it calls the Dragonfly portfolio that is built, in the company’s framing, for the agentic AI era. According to Qualcomm, the announcement centers on three new products: the Qualcomm Dragonfly C1000 CPU, Qualcomm High Bandwidth Compute (HBC), and the Qualcomm Dragonfly AI300 inference accelerator. The company also said that Qualcomm Technologies and Meta announced a strategic multi-generation collaboration for Qualcomm to supply data center CPUs to Meta.
The move pushes a company best known for smartphone processors and modems squarely into territory dominated by Nvidia, AMD, and Intel. CEO Cristiano Amon presented the effort as a data center infrastructure business rather than a single-chip entry, according to Jon Peddie Research.
What We Know
The Dragonfly C1000 is the centerpiece of Qualcomm’s CPU push. According to Qualcomm, it is a chiplet design with a 250+ core count running at frequencies above 5 GHz, with more than 2 TB/s of PCIe Gen 7 connectivity plus CXL connectivity. The company estimates the part will deliver more than 2x better performance per watt and expects commercial availability in 2028. The chiplet design is intended to deliver throughput and scale while preserving per-core performance, as the same release republished by Investing News Network describes it.
On the accelerator side, the AI300 is described by Qualcomm as a third-generation, air- and direct-liquid-cooled rack-level AI inference platform built around what it calls HBC Gen 2 technology, with the company expecting 4x to 8x better performance per watt compared to existing GPU-based architectures. Commercial sampling of the AI300 is expected in 2028. It joins the previously announced AI200 and AI250 accelerators, which Investing News Network notes were introduced the prior October, on an annual cadence roadmap.
High Bandwidth Compute is Qualcomm’s bid to rethink the memory bottleneck. The architecture combines accelerator logic with 3D-stacked DRAM in a tightly coupled package, according to Jon Peddie Research. Qualcomm says HBC is designed to enable a 6x increase in bandwidth per watt versus HBM, measured against competing published product specifications, per Qualcomm. The first HBC silicon is slated to ship inside the AI250 rack, which Qualcomm says is designed to enable 133 TB/s per card, an 18x increase in effective memory bandwidth, with commercial sampling expected in mid-2027.
Amon tied the roadmap directly to the shift toward AI agents. “Agentic AI is driving a significant increase in demand for AI inference in the data center. As these become the dominant workloads, infrastructure has to deliver much higher performance at lower power and cost,” said Cristiano Amon, President and CEO, Qualcomm Incorporated, in the Qualcomm release. “That plays directly to Qualcomm’s strengths, and we’re well positioned for this shift. With Qualcomm Dragonfly, we’re bringing our high-performance, low-power computing into the data center, with multi-year, multi-generation agreements with leading customers.”
The most concrete of those agreements is with Meta. Both Qualcomm and KuCoin report a multi-generation agreement under which Qualcomm Technologies will supply data center CPUs to Meta. Beyond the CPU deal, both Microsoft and Meta have identified themselves as early adopters of Dragonfly technologies, including HBC and Qualcomm server CPUs, according to Jon Peddie Research.
Qualcomm framed the opportunity in financial terms as well. The company projects roughly $300 million in data center revenue this fiscal year and $5 billion by fiscal 2027, according to Jon Peddie Research. It aims to grow data center revenue from roughly $0.3 billion in fiscal 2026 to more than $15 billion by fiscal 2029, per KuCoin. Amon expects the data center chip market to exceed $1 trillion by 2029 and wants Qualcomm to capture more than 5% of that opportunity, according to Jon Peddie Research.
What We Don’t Know
Much of the roadmap rests on hardware that has not yet shipped. The flagship C1000 CPU and AI300 accelerator are not expected to reach commercial availability or sampling until 2028, according to Qualcomm, leaving a multi-year gap between the announcement and meaningful product revenue. The scale and value of the Meta agreement were not disclosed in the materials, and the company’s published figures describe targets rather than booked orders. How HBC performs against entrenched HBM-based designs from Nvidia and AMD in real deployments also remains to be demonstrated.
Analysis
The Dragonfly launch reframes Qualcomm’s longstanding pitch — high performance at low power — for a market where energy and cooling, not raw peak throughput, increasingly govern total cost. By leading with HBC and a high-core-count CPU rather than a single flagship accelerator, Qualcomm is positioning itself as an infrastructure supplier across CPU, memory, and inference acceleration, the framing Amon emphasized per Jon Peddie Research. Securing Meta as a named CPU customer gives the roadmap an anchor tenant, but the long runway to 2028 silicon means investors and customers are underwriting a plan whose hardware is still years from shipping.