Qualcomm has officially extended its reach to the data center server market and launched the first CPU for this field-Dragonfly C1000. The chip isn't expected to enter mass production until the second half of 2028, and Meta will be one of the first customers to adopt it.
Dragonfly C1000 is Qualcomm's first server CPU designed specifically for data center workloads. It adopts a multi-chip design and self-developed Oryon core, with more than 250 cores and a maximum frequency of more than 5GHz. Qualcomm also positions it as a rack-level infrastructure component for the "intelligent AI" era, emphasizing its balance between performance and energy efficiency.
In terms of product planning, the mass production of Dragonfly C1000 is scheduled for the second half of 2028, and Meta will be the first to deploy this processor in its next-generation server cluster. Qualcomm also said that it has also received custom chip orders from two other undisclosed ultra-large-scale cloud customers, which means that its data center business is not just betting on a single partner.
From the technical specifications, this CPU is described as supporting LP DRAM memory subsystem, optional HBC connection, and being compatible with PCIe Gen 7 and CXL specifications. It also supports air cooling and liquid cooling solutions and is targeted at OCP ORv3 standard rack and server environments. These configurations indicate that Qualcomm hopes to make the Dragonfly C1000 a high-end server chip that can directly enter the infrastructure purchasing list of cloud vendors.
Qualcomm also announced a larger data center layout this time and announced the acquisition of AI infrastructure startup Modular to complete the software ecosystem. The market is still taking a wait-and-see attitude towards this acquisition and forward product route, because Dragonfly’s revenue will not be realized until 2028, and whether the related business can truly be transformed into stable growth depends on subsequent customer expansion and software ecosystem construction.