Qualcomm and AWS team up on custom silicon and optical connectivity for AI data centers

Qualcomm Technologies and Amazon have announced a multi-generational partnership to develop customized silicon for large-scale AI data centers, with a specific focus on AI inference workloads, alongside advanced optical connectivity solutions supporting speeds of up to 1.6T. The collaboration marks a significant step in the race to build more efficient and powerful infrastructure capable of handling the surging computational demands of generative AI and other large-scale machine learning models.

The partnership will combine Amazon's extensive AI infrastructure footprint with Qualcomm's chip design expertise. Under the agreement, the companies will jointly develop custom silicon tailored for AI inference, while the new optical connectivity work will draw on Qualcomm's SerDes and optical DSP technologies. Qualcomm also plans to expand its own use of AWS's AI infrastructure, including the Amazon Bedrock platform, for its electronic design automation workloads—a move that deepens the operational integration between the two firms.

"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 Incorporated. "Qualcomm is pleased to work with AWS on customized silicon and connectivity solutions, bringing decades of leadership in advanced processing and power-efficient compute, to deliver breakthrough performance and enable the next generation of AI infrastructure."

Prasad Kalyanaraman, vice president at AWS, echoed that sentiment. "This collaboration with Qualcomm Technologies builds on a strong foundation of partnership and reflects our shared commitment to pushing the boundaries of what's possible," he said. "By working together on customized silicon and advanced connectivity, we're delivering more performant, efficient, and cost-effective infrastructure for our customers."

AWS has been steadily expanding its in-house silicon portfolio, which includes the Trainium and Inferentia chips designed for AI workloads, as well as the Graviton CPU line. The Trainium3 generation, released in December 2025, is AWS's first 3nm AI chip; the company says the Trainium3 UltraServer delivers up to 4.4 times more compute performance, four times greater energy efficiency, and nearly four times more memory bandwidth than its predecessor. Qualcomm, meanwhile, unveiled several new data center products in June of this year, including the Dragonfly C1000 CPU, the High Bandwidth Compute (HBC) platform, and the Dragonfly AI300 inference accelerator. Meta has already signed on as a customer for the Dragonfly C1000 and plans to deploy the chips in its next-generation server fleet.

The deal underscores a broader industry shift toward purpose-built, vertically integrated AI infrastructure. As hyperscale operators and their suppliers race to balance performance with power efficiency, partnerships that span chip design, connectivity, and cloud services are becoming increasingly central to maintaining competitive advantage. By aligning Qualcomm's low-power compute heritage with AWS's scale and cloud-native AI services, the two companies are positioning themselves to address one of the most pressing challenges in modern data center operations: delivering exponentially greater compute capacity without proportional increases in energy consumption.