# Amazon and Qualcomm Join Forces to Shape the Future of AI Data Centre Silicon
A landmark partnership between two of the technology industry’s biggest players is set to reshape the landscape of artificial intelligence infrastructure. Amazon and Qualcomm have entered into a multi-generational agreement to design and develop custom semiconductor products specifically tailored for AI data centres, signalling a deepening of their technical collaboration that spans both processing power and networking capabilities.
## A New Era of Custom Silicon for Cloud AI
Under the terms of the agreement, Qualcomm will work alongside Amazon to produce specialised processors optimised for AI inference workloads. The partnership also encompasses the development of cutting-edge optical connectivity solutions capable of reaching speeds of 1.6 terabits per second, with both companies planning for future iterations of this technology as demand grows.
This development positions Qualcomm as a new addition to the broad ecosystem of silicon providers that support Amazon Web Services’ AI operations. AWS has long relied on a diverse mix of processors, including its own proprietary Trainium and Inferentia accelerator chips, as well as instances powered by graphics processors from Nvidia. The inclusion of Qualcomm-designed silicon suggests that Amazon is actively seeking to diversify its hardware supply chain and reduce dependence on any single vendor.
## What Does the Collaboration Cover?
The scope of the Amazon-Qualcomm partnership extends across several critical areas of data centre technology:
– **Custom processors** designed for AI inference workloads, with no public details yet released about architecture, manufacturing process, or launch timelines.
– **Optical connectivity** leveraging Qualcomm’s SerDes and optical digital signal processor technologies, enabling high-bandwidth internal links within Amazon’s data centre facilities.
– **Future-generation silicon**, with both companies committed to developing successive iterations of their joint products over time.
Neither company has publicly disclosed whether the custom Qualcomm processors will be made available as standalone AWS instances for customers, or whether they will be used exclusively within Amazon’s own data centre operations. Similarly, it remains unclear how these processors relate to Qualcomm’s existing Dragonfly data centre portfolio, which includes the C1000 CPU, AI200, AI250, and AI300 inference accelerators, as well as dedicated networking products.
## Networking and Connectivity at Scale
One of the most technically ambitious aspects of the collaboration is the push toward 1.6 terabit optical connectivity. Qualcomm brings to the table a well-established portfolio of high-speed connectivity technologies, including support for 800G and 1.6T links across optical modules and various interconnect solutions. The integration of these technologies into Amazon’s data centre architecture could significantly boost the speed and efficiency of data movement between servers, storage systems, and networking equipment.
This push in networking is underpinned by Qualcomm’s broader strategic expansion. In December 2025, the company completed its acquisition of Alphawave Semi for an implied enterprise value of approximately $2.4 billion. The deal brought Alphawave’s expertise in high-speed wired connectivity, custom silicon design, and chiplet technologies into Qualcomm’s data centre division. Tony Pialis, co-founder and former CEO of Alphawave, subsequently assumed leadership of Qualcomm’s data centre operations, further consolidating the company’s ambitions in this space.
## Financial Terms and Strategic Commitments
Details of the commercial arrangement were disclosed through a filing with the United States Securities and Exchange Commission. According to the filing, Qualcomm issued an affiliate of Amazon a warrant that allows the purchase of up to 25 million Qualcomm shares at an exercise price of $161.26 per share. The shares vest in stages, contingent on commercial agreements, binding purchase orders, and the purchase of Qualcomm server chips, technology, systems, and manufacturing services.
The maximum value of payments linked to the warrant’s vesting conditions stands at $60 billion, though the filing clarifies that Amazon has not necessarily committed to spending that amount. At the time of the warrant’s issuance, 3.75 million shares were vested based on initial purchase commitments. The warrant is set to expire in September 2036, giving the partnership a lengthy runway over which to develop and commercialise its joint products.
## Qualcomm’s Broader Data Centre Ambitions
The Amazon collaboration is part of Qualcomm’s wider push into the data centre market. At its investor day in June 2026, the company set an ambitious target of generating more than $15 billion in data centre revenue by fiscal 2029. The Dragonfly product roadmap forms the cornerstone of this strategy, combining CPUs, AI accelerators, and high-speed networking components into a unified offering that follows an annual product release cadence.
Beyond standard product lines, Qualcomm is also investing in custom silicon solutions for individual clients, a model that mirrors approaches adopted by other major technology companies. The September announcement regarding the Amazon partnership did not reveal specifics about the processors’ architecture, memory configuration, power requirements, or expected availability.
## The Broader Industry Context
Amazon is not alone in pursuing a diversified AI hardware strategy. Other hyperscale technology companies are building their own specialised processors alongside third-party options. Microsoft operates its Maia accelerators alongside hardware from Nvidia and AMD. Google Cloud offers its Tensor Processing Units (TPUs) as a distinct alternative to GPUs and other accelerators. Meanwhile, Meta continues to develop its MTIA processors for internal workloads, with the MTIA 300, introduced in August, designed specifically for training ranking and recommendation models and featuring integrated networking components.
This industry-wide trend toward custom silicon reflects a growing recognition that off-the-shelf hardware cannot always meet the unique demands of large-scale AI workloads. By developing purpose-built processors, companies can optimise for performance, power efficiency, and cost in ways that generic solutions cannot match.
## Frequently Asked Questions (FAQ)
**Q: What is the Amazon-Qualcomm partnership about?**
A: Amazon and Qualcomm are collaborating to develop multiple generations of custom silicon designed for AI data centres, including processors for AI inference and optical connectivity solutions capable of speeds up to 1.6 terabits per second.
**Q: Will the Qualcomm processors be available as AWS instances?**
A: Neither company has disclosed whether the custom processors will be offered as AWS instances for customers or used internally within Amazon’s data centres.
**Q: How does this affect Amazon’s relationship with Nvidia?**
A: Amazon has not indicated that the Qualcomm collaboration is intended to replace or reduce its reliance on Nvidia hardware. AWS continues to deploy Blackwell-based EC2 instances across multiple regions alongside its own Trainium and Inferentia processors.
**Q: What is the Dragonfly portfolio?**
A: Dragonfly is Qualcomm’s data centre product family, which includes the C1000 CPU, AI200, AI250, and AI300 inference accelerators, and high-speed networking and connectivity products.
**Q: What financial terms underpin the collaboration?**
A: Qualcomm issued an Amazon affiliate a warrant for up to 25 million shares at $161.26 per share. Up to $60 billion in linked payments could trigger full vesting, though Amazon has not confirmed spending that amount. The warrant expires in September 2036.
**Q: How does this fit into the wider AI hardware landscape?**
A: The partnership reflects a broader industry trend in which major cloud providers are developing or sourcing custom silicon — including Microsoft’s Maia, Google’s TPUs, and Meta’s MTIA processors — to optimise AI infrastructure for their specific workloads.
**Q: When will the custom processors be available?**
A: No deployment dates have been announced for either the custom processors or the 1.6 terabit connectivity technology.
**Q: Why is optical connectivity important for AI data centres?**
A: High-speed optical connectivity enables rapid data transfer between servers and components within a data centre, which is critical for AI workloads that involve massive datasets and distributed computing tasks.
## Conclusion
The Amazon-Qualcomm partnership represents a significant step forward in the evolution of AI data centre infrastructure. By combining Amazon’s massive scale and cloud expertise with Qualcomm’s growing capabilities in custom silicon, AI acceleration, and high-speed connectivity, the collaboration has the potential to introduce new hardware options that could enhance performance, efficiency, and flexibility for AI workloads at scale.
While many details remain undisclosed — including the architecture of the custom processors, their relationship to existing Qualcomm products, and specific deployment timelines — the strategic intent is clear: both companies are investing in a long-term relationship that could reshape how AI infrastructure is built and operated.
As the broader technology industry continues its race to develop specialised AI hardware, partnerships like this one underscore a fundamental shift: the era of one-size-fits-all processors is giving way to a new landscape of highly tailored, application-specific silicon designed to meet the exacting demands of modern artificial intelligence.
Thank you for reading
