퀄컴은 아마존과 다세대 AI 칩 계약을 체결하며, 차세대 실리콘 세대에 걸친 장기 도입 권리를 확보했다.
Qualcomm and Amazon have entered a multi-generation agreement to develop customized silicon for AWS data centers, beginning with chips optimized for AI inference. As enterprises increasingly deploy AI workloads into production, AWS is expanding its internal semiconductor strategy. Qualcomm’s participation will provide Amazon with greater architectural flexibility and improved cost management for its AI services. Beyond processor development, the partnership will also address escalating bandwidth demands through next-generation optical interconnect solutions.
The financial architecture of the agreement ties Amazon’s potential equity stake directly to commercial execution. Amazon has received warrants to acquire 25 million Qualcomm shares at $161.26 per share, representing approximately $4 billion in value if fully exercised. These warrants vest incrementally based on commercial milestones, including Amazon’s procurement commitments of up to $60 billion in Qualcomm server chips and related technologies. The warrants are scheduled to expire in September 2036.
Initial development efforts will prioritize AI inference—the computational phase where deployed models process incoming requests and generate outputs. As AI applications scale, this workload shifts deeper into data centers, intensifying demands on compute performance, energy efficiency, and operational expenditures. To support these workloads, Qualcomm will integrate its SerDes and optical digital signal processing (DSP) technology into connectivity architectures capable of reaching 1.6 terabits per second, with subsequent generations designed to accommodate AWS’s ongoing bandwidth requirements.
This partnership marks a strategic expansion for Qualcomm, traditionally known for mobile processors, as it targets $15 billion in annual data center revenue by fiscal year 2029. The company’s Dragonfly C1000 CPU, unveiled in June for data center and agentic AI workloads, has already secured Meta as a launch customer ahead of 2028 production ramp. Qualcomm’s broader semiconductor roadmap includes dedicated AI accelerators and multi-chip interconnect technologies, positioning the firm to compete directly with NVIDIA in GPUs and Intel and AMD in CPUs.
The collaboration extends beyond hardware supply into mutual infrastructure integration. Qualcomm plans to leverage AWS’s AI platforms, including Amazon Bedrock, to accelerate electronic design automation workloads, significantly reducing chip design cycles. By developing custom silicon for AWS while simultaneously utilizing Amazon’s cloud-based AI infrastructure to engineer future semiconductor products, both companies are establishing a tightly coupled feedback loop between cloud operations and chip development.