Amazon and Qualcomm have entered a long-term collaboration centered on customized silicon and high-speed connectivity for artificial intelligence data centers, opening a potentially significant new line of business for a chipmaker still best known for smartphones.
The companies plan to collaborate across multiple product generations, combining Amazon Web Services’ AI infrastructure with Qualcomm’s experience in silicon design, power-efficient computing and system integration. Their work will focus on customized chips for AI inference and high-speed optical connectivity for Amazon’s data-center networks.
The agreement broadens Amazon’s supplier base as cloud operators search for more control over the cost, performance and energy demands of AI computing. It also gives Qualcomm an opportunity to establish itself in a data-center market where Nvidia remains the leading provider of AI accelerators.
A Partnership Covering Compute and Connectivity
the companies will develop customized silicon at scale for AI inference. Inference is the stage at which a trained model processes new information and produces an output, making it central to commercial AI services that must respond quickly to large numbers of requests.
The work will also extend beyond processors. Qualcomm said the companies are developing optical-connectivity technology capable of reaching 1.6 terabits per second, alongside future generations of connectivity products. These components move data between servers, chips and other parts of an AI cluster, where insufficient bandwidth can limit the performance of otherwise powerful computing systems.
Cristiano Amon, president and CEO, Qualcomm Incorporated, said: “As AI demand accelerates, data center infrastructure will require advances in both computing and connectivity to deliver greater performance with more efficiency.”
Amazon already designs its own processors, including its Trainium family of AI chips. Bringing Qualcomm into the development process does not necessarily signal a retreat from that strategy. Instead, the partnership gives Amazon access to another pool of engineering and connectivity expertise as it expands its custom-silicon operations and reduces its reliance on any single outside supplier.
Prasad Kalyanaraman, vice president, AWS, stated: “By working together on customized silicon and advanced connectivity, we’re delivering more performant, efficient, and cost-effective infrastructure for our customers.”
Qualcomm will deepen its own use of AWS infrastructure under the arrangement. The company plans to use services including Amazon Bedrock for electronic-design-automation workloads, with the aim of shortening chip-development cycles.
A $60 Billion Ceiling is Tied to Purchases
Amazon could purchase up to $60 billion in Qualcomm AI data-center chips and related products under the long-term arrangement, based on a regulatory filing connected to the transaction, according to a report by Reuters.
Qualcomm also granted Amazon warrants allowing it to acquire as many as 25 million Qualcomm shares at an exercise price of $161.26 per share. Exercising all the warrants at that price would represent an investment of approximately $4.03 billion. Their vesting is linked to purchasing milestones, meaning Amazon would earn the right to acquire more shares as qualifying orders increase.
The largest purchasing milestone reaches $60 billion, but the structure should not be presented as a firm $60 billion order. Amazon is not necessarily required to spend the full amount, and Qualcomm has not said that the maximum will be reached. Describing the agreement as being “worth $60 billion” without explaining its conditional nature could therefore overstate the commitment.
Qualcomm Expands Beyond Mobile Chips
Data centers have become an important part of Qualcomm’s effort to reduce its dependence on smartphone processors and wireless modems. The company has been building a broader portfolio spanning AI compute, custom silicon and the high-speed connections required to link large numbers of processors inside modern server systems.
Its acquisition of Alphawave Semi added optical and wired-connectivity technology to that strategy. Qualcomm agreed to acquire the British semiconductor company in a transaction valued at approximately $2.4 billion, seeking capabilities in areas including custom silicon, connectivity products and chiplets.
Competition will remain intense. Nvidia has established a large AI-computing ecosystem built around its accelerators, networking products and software, while other semiconductor companies are pursuing custom-chip contracts with major cloud providers. Qualcomm’s work with Amazon gives it a substantial customer relationship.
Amazon, meanwhile, gains another route for tailoring infrastructure to its own workloads. Custom designs can help cloud providers balance processing speed, power consumption and cost more closely than general-purpose hardware, but developing them across several generations requires significant capital and long execution cycles.
