MPN: Dragonfly C1000
Qualcomm Dragonfly C1000
250+ Cores · Oryon · PCIe Gen 7

Custom Qualcomm Oryon cores above 5GHz in a 250+ core chiplet design · >2× performance per watt versus competitor server CPUs · >2TB/s PCIe Gen 7 with CXL · OCP ORv3 compliant

Agentic CPU

Purpose-built for high-throughput agentic orchestration and low-latency interactive AI serving.

General-purpose CPU

Optimised for performance-per-TCO in first-party fleets and performance-per-vCPU in third-party elasticity.

AI head node CPU

Maximises XPU utilisation for large-scale generative AI compute clusters.

Dragonfly Platform

Rack-scale unified platform bringing compute, AI, memory and connectivity together for agent-driven workloads.

Specifications

Vendor

Qualcomm Technologies, Inc.

Product

Qualcomm Dragonfly C1000

Type

Data center CPU (server processor)

CPU Cores

250+ cores (chiplet design)

Architecture

Custom Qualcomm Oryon cores

Frequency

Above 5 GHz

Performance / Watt

>2× vs competitor server CPUs

Interconnect

>2 TB/s PCIe Gen 7 + CXL

Memory

Low-power memory subsystem (bandwidth, capacity, latency, power)

Variants

Agentic · General-purpose · AI head node

Inference

CPU-based inference with optional HBC attach

Cooling

Air and liquid cooling

Compliance

OCP ORv3 compliant racks and servers

RAS

ECC, fault isolation, error recovery

Availability

Commercial availability expected 2028

Stock

Available

Price

Quote

Overview

Qualcomm unveiled the Dragonfly C1000 at its Investor Day on June 24, 2026, marking the company's entry into the data center CPU market as the centrepiece of the wider Qualcomm Dragonfly portfolio spanning CPUs, AI inference accelerators, memory architecture and connectivity.

The C1000 pairs custom Qualcomm Oryon CPU cores running above 5GHz with a 250+ core chiplet design, delivering more than 2× better performance per watt than competing server CPUs. Connectivity is equally aggressive: over 2 TB/s of PCIe Gen 7 plus CXL for accelerators, networking, storage and memory disaggregation, backed by a low-power memory subsystem tuned for bandwidth, capacity and latency.

Qualcomm positions the C1000 squarely at agentic AI infrastructure and offers it in three distinct configurations — an agentic CPU, a general-purpose CPU and an AI head node CPU — each tuned to a different role in the AI factory. The part supports air and liquid cooling, is OCP ORv3 compliant, and includes enterprise RAS features (ECC, fault isolation, error recovery). Commercial availability is expected in 2028, with Qualcomm naming Meta as a launch customer.

Need Qualcomm Dragonfly C1000?

Contact QS Compute for availability, configuration, and volume pricing.

Request Quote