AWS is using Qualcomm for AI inference while Qualcomm uses AWS Bedrock to design the chips
decoder ·
Qualcomm is designing custom AI inference chips for AWS across multiple product generations, and the companies are also collaborating on optical interconnects capable of up to 1.6 terabits per second. Qualcomm brings its expertise in power-efficient chip design, while Amazon provides its cloud infrastructure for the partnership. In return, Qualcomm uses Amazon Web Services, including Amazon Bedrock, to accelerate its own chip design process. This deal marks Amazon as Qualcomm's third major data center win since June, following Meta's adoption of the Dragonfly C1000 processor and Microsoft's use of Qualcomm's HBC memory architecture in Azure. Qualcomm has set a goal of reaching $15 billion in data center revenue by 2029. AWS will integrate Qualcomm's designs alongside its own Trainium, Graviton, and Nitro chip families, targeting inference workloads where energy cost per token is a key factor. Efficiency is also Qualcomm's focus on edge devices, with its AI research division releasing a framework in March to run reasoning models on smartphones.
decoder