Qualcomm and AWS Partner on AI Silicon and 1.6T Connectivity

Qualcomm and Amazon Web Services (AWS) have launched a multi-generation collaboration to develop customized silicon and high-performance connectivity for AI data centers. The deal signals a strategic alignment between the chipmaker and the cloud giant to optimize the hardware powering large-scale AI inference.

The partnership focuses on the creation of tailored AI inference chips and optical connectivity solutions capable of supporting speeds up to 1.6 terabits per second (1.6T). The financial structure of the agreement is substantial: Qualcomm has issued a warrant to Amazon for up to 25 million shares at an exercise price of $161.26 per share, representing a maximum warrant value of approximately $4 billion. According to reports, 3.75 million of these shares vested immediately based on Amazon's initial purchase commitments. The remaining shares are tied to commercial milestones involving payments to Qualcomm totaling up to $60 billion.

Diversifying Beyond Mobile

For Qualcomm, this move is a critical step in its aggressive diversification strategy. Long regarded as the dominant force in smartphone and mobile processors, the company is now pivoting toward data center infrastructure. This transition is already underway; Qualcomm recently unveiled the Dragonfly C1000 CPU, a chip designed for the data center. Industry reports indicate that Meta plans to deploy the C1000 in production by 2028, further cementing Qualcomm's footprint in the server market.

The Impact on AI Infrastructure

This collaboration addresses the growing bottleneck in AI scaling: the need for both compute power and the connectivity to move data between chips. "As AI demand accelerates, data center infrastructure will require advances in both computing and connectivity to deliver greater performance with more efficiency," said Cristiano Amon, President and CEO of Qualcomm.

For Amazon, the deal reduces its reliance on generic, off-the-shelf silicon. By co-developing customized hardware, AWS can deploy power-efficient chips specifically optimized for the massive inference workloads required by its cloud customers. Prasad Kalyanaraman, Vice President at AWS, noted that the joint effort in customized silicon and advanced connectivity is intended to deliver more performant and cost-effective infrastructure.

Future Outlook

As the industry shifts from training massive models to deploying them at scale, the efficiency of inference hardware will become a primary competitive advantage. Market observers will now watch for the specific rollout of the 1.6T connectivity solutions and the pace at which Amazon hits the commercial milestones required to vest the remaining Qualcomm shares. While the partnership establishes a long-term roadmap, the actual performance gains of these customized chips compared to existing GPU-based solutions remain the key metric for the next phase of deployment.