Qualcomm Secures Major Meta Data Center Chip Deal, Officially Entering the AI Server Track

On June 24 local time, Qualcomm announced a landmark strategic cooperation to supply its self-developed Dragonfly C1000 data center processors for large-scale deployment in Meta’s proprietary AI data centers, with the chip officially scheduled for launch in 2028. The partnership marks Qualcomm’s formal entry into the core AI server market long dominated by NVIDIA, AMD and Intel, after years of focusing on mobile chip development. Meanwhile, Qualcomm’s CFO set a clear performance target, aiming to push the annual revenue of its AI data center business beyond $5 billion in the 2027 fiscal year and build a robust new growth driver for the company’s trillion-dollar annual revenue goal.

For a long time, Qualcomm’s revenue has relied heavily on smartphone and automotive chips, with solid technological barriers in low-power mobile computing. However, the company has long been absent from the high-value data center computing market. With the rapid popularization of generative AI, global cloud computing demand has surged. Unlike power-hungry and costly AI training computing, scenarios such as large model inference, content recommendation and AI agent operation require high-efficiency and low-cost computing power. This has enabled Qualcomm’s long-accumulated advantages in low-power chip architecture, self-developed cores and high-speed interconnection technology to find vital new application scenarios.

The core product of this cooperation, the Dragonfly C1000, is a flagship server processor exclusively built for AI inference workloads. Adopting Qualcomm’s self-developed Oryon cores and advanced Chiplet heterogeneous architecture, it delivers balanced high performance and ultra-low power consumption. Compared with traditional x86 server chips, the new processor doubles energy efficiency, perfectly adapting to Meta’s core businesses including multimodal large model inference and social algorithm scheduling. Supporting diversified data center deployment solutions, it meets the large-scale cluster operation needs of super-scale data centers and precisely addresses cloud vendors’ core demands for cost reduction and efficiency improvement.

Meta’s partnership with Qualcomm is a crucial layout of its diversified computing supply chain strategy. As one of the world’s top AI computing purchasers, Meta has historically relied heavily on NVIDIA GPU solutions, while supplementing its demand with self-developed MTIA inference chips. Nevertheless, in-house chips feature long R&D cycles and high ecosystem adaptation costs, and reliance on a single supplier brings prominent supply chain risks. To optimize its layout, Meta continues to adopt NVIDIA GPUs for high-end AI training, while introducing Qualcomm’s ARM-based chips to undertake massive inference computing tasks. This multi-vendor hardware combination helps cut operational costs and mitigate supply risks. Far from a one-time purchase, the cooperation is a long-term binding agreement covering multi-generation chip iterations, providing Qualcomm with sustainable and stable order support.

This high-profile deal with a leading client serves as a milestone for Qualcomm’s transformation into the cloud computing track. Previously, Qualcomm’s data center chip business remained in the small-scale trial stage without core orders from top-tier cloud vendors. The in-depth cooperation with Meta not only verifies the product competitiveness of the Dragonfly series but also drives multiple global leading cloud providers to finalize procurement intentions, rapidly expanding Qualcomm’s cloud client ecosystem. To shore up its cloud software shortcomings, Qualcomm has actively optimized its ecological layout and acquired AI infrastructure enterprises to achieve seamless compatibility between self-developed chips and large model development frameworks. By building a full-stack computing solution, it has formed differentiated competitiveness against NVIDIA’s CUDA ecosystem.

The global AI server chip market is undergoing accelerated structural restructuring, with ARM architecture emerging as the core growth driver. Traditional x86 chips suffer from high power consumption and excessive operation and maintenance costs, making them unsuitable for massive inference scenarios. In contrast, ARM-based products including AWS Graviton, NVIDIA Grace and Qualcomm Dragonfly are rapidly capturing market share by virtue of outstanding energy efficiency. Industry forecasts indicate that the global server CPU market will exceed $170 billion by 2030, with AI inference scenarios contributing over 60% of the incremental growth, and ARM architecture shipment share will rise significantly, allowing Qualcomm to precisely seize the industry’s development dividend.

Nevertheless, Qualcomm still faces multiple challenges in its cross-border expansion. First, the Dragonfly C1000 will not be commercially available until 2028, during which competitors will launch iterative products and intensify market competition. Second, the ARM cloud ecosystem is less mature than the traditional x86 system, bringing certain adaptation costs for client architecture migration. In addition, NVIDIA maintains a monopoly over the high-end AI training market with its comprehensive software and hardware ecosystem, limiting Qualcomm’s business growth to the segmented inference track in the short term.

Overall, the strategic cooperation between Qualcomm and Meta breaks the inherent pattern of the AI server chip industry and promotes the diversification of the global computing supply chain while reducing over-reliance on single vendors. For Qualcomm, this entry eliminates its long-term over-reliance on mobile chip businesses and completes a pivotal transformation from a terminal chip manufacturer to a full-stack computing service provider. The official mass production of the Dragonfly C1000 in 2028 will trigger the large-scale growth of Qualcomm’s cloud AI business. Behind the $5 billion revenue target lies a trillion-level market growth space, which will further popularize low-power ARM computing in data center construction and reshape the competitive landscape of the global AI chip industry.