Qualcomm has struck a deal with Amazon to develop customised processors for large-scale artificial intelligence (AI) data centre infrastructure, in an agreement linked to up to $60 billion in potential commercial transactions.
Under the arrangement, Qualcomm has also issued Amazon a warrant to acquire up to 25 million shares in the chipmaker at $161.26 each, giving the technology company a potential stake worth up to $4 billion.
The partnership will focus on developing processors optimised for AI inference workloads across Amazon Web Services (AWS) data centres, combining Qualcomm’s power-efficient compute technology with AWS’ global cloud infrastructure.
The companies will also work on high-performance optical connectivity technologies to improve communication between large clusters of AI chips.
Qualcomm plans to supply optical digital signal processors and serializer/deserializer products capable of supporting speeds of up to 1.6 Tb/s. The technologies are intended to reduce data bottlenecks in data centre networks and enable AI chips to communicate with minimal latency.
The agreement represents Qualcomm’s latest expansion into the data centre sector as it seeks to broaden its business beyond smartphones. The company has recently pursued an AI-native foundation, acquired AI software specialist Modular and advanced its agentic AI ambitions beyond edge devices.
The collaboration will also see Qualcomm deepen its use of AWS infrastructure. The chipmaker plans to use cloud-based AI tools, including Amazon Bedrock, for electronic design automation workloads aimed at accelerating its chip development cycles.
Qualcomm CEO Cristiano Amon said the rapid growth in AI demand required advances in both processing and network interconnect technologies to improve operational efficiency.
AWS Vice-President Prasad Kalyanaraman said the co-development effort would deliver more performant and cost-effective infrastructure for cloud customers globally.
For AWS, the partnership with Qualcomm provides another route to addressing the high power consumption and rising costs associated with operating AI services at scale.
