en.Wedoany.com Reported - Qualcomm Technologies, Inc. announced a series of new data center products at its Investor Day, including the Qualcomm Dragonfly C1000 CPU, Qualcomm High Bandwidth Computing (HBC), the Qualcomm Dragonfly AI300 inference accelerator and its connectivity products, as well as custom silicon solutions. The entire portfolio is designed around the goal of "maximizing performance per watt and token throughput at a lower total cost of ownership."

These new platforms mark Qualcomm Technologies' accelerated efforts to build an AI-optimized, full-stack data center infrastructure, encompassing agent-level and data center-level CPUs, AI inference accelerators, high-performance connectivity, and large-scale custom silicon solutions. Among them, the Dragonfly AI300, together with the previously announced Dragonfly AI200 and AI250, forms the data center product portfolio, accompanied by an AI accelerator roadmap updated on an annual cadence.

"Agentic AI is driving a significant surge in demand for AI inference in data centers," said Cristiano Amon, President and CEO of Qualcomm Incorporated. "As these workloads become mainstream, infrastructure needs to deliver higher performance at lower power and cost. This is precisely where Qualcomm excels, and we are well-prepared. With the Dragonfly series, we bring high-performance, low-power computing to the data center and have signed multi-year, multi-generation agreements with key customers."

In building a reasoning-first platform for hyperscale data centers, Qualcomm Technologies leverages its expertise in system-on-chip (SoC), low-power design, high-performance processing, and leading IP, combined with engineering experience from over 40 billion components, to introduce a disaggregated rack-level AI infrastructure. This platform is specifically designed for agent-intensive AI inference workloads in hyperscale data center environments, enabling better token economics, low latency, simplified integration, scalable deployment, and lower total cost of ownership. Facing the surge in token demand driven by agentic AI, Qualcomm Technologies focuses on "tokens per watt" as an optimization lever to reduce total cost of ownership (TCO).

"What enterprises need today goes far beyond individual components," said Tony Pialis, Executive Vice President and General Manager of Data Center at Qualcomm Technologies, Inc. "Coordinating multiple types of computing across distributed, always-on infrastructure is critical. With Qualcomm Dragonfly, we integrate compute, AI, memory, and connectivity into a unified rack-level platform, purpose-built for increasingly complex agent-driven workloads while addressing bottlenecks in memory bandwidth and power consumption. This builds on Qualcomm Technologies' core capabilities over many years—high-performance, low-power computing at scale, now applied to the data center in a way few companies can match."