Qualcomm CEO confirms Dragonfly C1000 CPU on track for Meta

Qualcomm is back in the server CPU business, and it wants you to know the train is running on time. The company’s CEO has confirmed that the Dragonfly C1000, its ambitious data center processor built for Meta’s server infrastructure, remains on schedule for delivery as planned.

What the Dragonfly C1000 actually is

Qualcomm announced the Dragonfly C1000 on June 24, 2026, positioning it as a high-performance data center CPU built for agentic AI and general-purpose server workloads.

The chip uses a chiplet design and houses over 250 cores, powered by custom Oryon CPU cores capable of exceeding 5 GHz. For context: chiplet architecture means the processor is assembled from multiple smaller dies rather than one giant piece of silicon, which makes it easier to push core counts higher while keeping manufacturing yields manageable.

Qualcomm also claims the C1000 delivers more than two times the performance per watt compared to competing server CPUs. That efficiency claim is the core of Qualcomm’s pitch: it spent years perfecting power-efficient processing in mobile chips, and now it’s arguing that same discipline translates directly to data centers, where electricity costs are a serious operational concern.

The chip also supports PCIe Gen 7, the latest generation of the interconnect standard that governs how fast data moves between processors and other components inside a server. And the broader Dragonfly platform includes a High Bandwidth Compute architecture, with sampling expected in the first half of 2027. Full production for Meta’s servers is targeted for the second half of 2028.

The Meta partnership and what it means for Qualcomm

Qualcomm’s arrangement with Meta is a multi-generation agreement, not a one-off order. Meta intends to use the C1000 to expand its server fleets and, according to the company, support its broader push toward what it describes as personal superintelligence.

For Qualcomm, the Meta deal functions as a highly visible anchor customer. When you’re trying to re-enter a market you abandoned years ago, having the world’s largest social network publicly committed to your roadmap is about as credible a reference as you can get.

Qualcomm exited the server CPU market in 2018, so this return has been a long time coming. The company sold off its Centriq server CPU business that year after struggling to gain traction against Intel’s entrenched position and AMD’s resurgent Epyc line.

The internal target Qualcomm has set for this segment is approximately $15 billion in annual data center sales by fiscal 2029. That’s an aggressive number for a company that had essentially zero revenue in this category for most of the past decade.

Why the competitive landscape makes this interesting

The server CPU market Qualcomm is re-entering looks very different from the one it left. AMD has made serious inroads against Intel with its Epyc processors, and Arm-based designs from Amazon’s Graviton line and Ampere Computing have demonstrated that x86 isn’t the only viable architecture at scale.

Qualcomm’s Oryon cores are also Arm-based, which puts it in a growing camp rather than a shrinking one. The Arm server ecosystem has matured considerably since Qualcomm’s first attempt, with better software support and more cloud operators willing to run workloads on non-x86 hardware.

The power efficiency angle is also more strategically timed than it might have been five years ago. Data centers are consuming electricity at a rate that is becoming a genuine policy issue in multiple countries. If Qualcomm’s claim of more than two times the performance per watt holds up under real-world conditions, that’s not just a technical spec. It’s a cost argument and an energy argument that lands differently in 2026 than it would have in 2016.