Qualcomm and Amazon Forge $60 Billion Custom Silicon Deal for AI Data Centers

21 Sources

Share

Qualcomm has struck a multi-generational partnership with Amazon to develop custom silicon for AI inference and 1.6 Tbps optical networking technology. The deal includes $4 billion in warrants and could total up to $60 billion in commercial commitments through 2036, marking Qualcomm's major push into data center infrastructure.

Qualcomm Amazon Partnership Unlocks Major AI Data Centers Expansion

Qualcomm has announced a sweeping multi-generational custom silicon agreement with Amazon Web Services that could reshape the AI data center infrastructure landscape. The partnership, unveiled on Tuesday, involves developing customized chips for AI inference and high-performance optical connectivity solutions extending to 1.6 Tbps.

1

Amazon's purchases and commercial commitments under this arrangement could reach up to $60 billion through September 2036.

1

Source: TechRadar

Source: TechRadar

As part of the deal structure, Qualcomm issued warrants allowing Amazon to acquire 25 million shares at $161.26 each, representing a $4 billion position.

3

These warrants vest in tranches tied to commercial milestones and server chip purchases, expiring on September 3, 2036.

3

Qualcomm shares surged more than 9% in premarket trading following the announcement, offering relief to investors after the stock slipped about 1% earlier this year due to weaker smartphone demand.

2

Custom Chips for AI Data Centers Target Inference Workloads

The collaboration centers on developing custom silicon specifically optimized for AI inference, a fast-growing market focused on running trained AI models that has become a key battleground among semiconductor firms.

2

While details remain sparse, the arrangement could involve fully custom silicon, chiplets, or customized versions of Qualcomm's emerging datacenter accelerator portfolio.

1

Source: Interesting Engineering

Source: Interesting Engineering

Qualcomm's AI250-series Dragonfly rack systems, expected to launch next year, employ an innovative approach to address memory bandwidth bottlenecks in AI workloads. The technology swaps expensive high-bandwidth memory for what Qualcomm calls "high-bandwidth compute," achieving higher effective memory bandwidth using cheaper LPDDR memory by shifting compute to the memory's base die.

1

This approach mirrors how Intel builds customized Xeon 6 processors for AWS, suggesting Qualcomm may deliver tailored versions of its AI250-series chips to the cloud provider.

1

Qualcomm is also developing the C1000 datacenter CPU, which the company claims will reach 5 GHz and scale beyond 250 cores when it arrives in the second half of 2028.

1

The chipmaker previously unveiled the Dragonfly C1000 data center processor in June with Meta as a customer from 2028, setting a $15 billion data center revenue target for fiscal 2029.

3

High-Performance Optical Connectivity Addresses AI Infrastructure Bottlenecks

Beyond custom silicon, the AI infrastructure collaboration tackles a critical challenge facing modern AI data centers: moving massive volumes of data between thousands of interconnected computing devices. Qualcomm will supply serializer-deserializers and optical digital signal processors to boost port speeds to 1.6 Tbps, likely leveraging technology acquired through the purchase of Alphawave Semi late last year.

1

High-speed optical interconnects have emerged as a major supply chain hurdle for scaling AI infrastructure, particularly as compute expands from racks with a few dozen accelerators to hundreds or thousands spanning multiple racks.

1

The partnership points to AWS expanding its supply chain to include Qualcomm as a source of optical pluggables, or potentially near- or co-packaged optics.

1

"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.

4

AWS VP Prasad Kalyanaraman added, "By working together on customized silicon and advanced connectivity, we're delivering more performant, efficient, and cost-effective infrastructure for our customers."

5

Source: The Register

Source: The Register

Strategic Implications for Power Efficiency and Market Position

The collaboration brings together Amazon's comprehensive AI infrastructure with Qualcomm's expertise in power-efficient processing, silicon design, and system-level integration.

4

This focus on power efficiency carries significant weight as data centers face mounting pressure to reduce resource consumption. In Amazon's 2025 sustainability report, the company declared a global average Power Usage Effectiveness of 1.14, nearly 9% better than the public cloud industry average of 1.25, though still trailing Google's fleet-wide average of 1.09.

5

The customer-supplier relationship operates bidirectionally. While AWS acquires Qualcomm's chip technology, Qualcomm plans to expand its use of Amazon Bedrock platform for electronic design automation workloads, targeting a reduction in chip design cycles.

4

Bank of America expects the server processor market to more than double from $27 billion in 2025 to $60 billion by 2030, with Nvidia, Intel, and AMD all competing aggressively.

3

Qualcomm's commitment to AWS arrives as the chipmaker pursues broader data center ambitions. The company was in talks to acquire Modular, an AI infrastructure software company, for approximately $4 billion in June.

3

Qualcomm is also one of three chipmakers, alongside Nvidia and AMD, that signed memoranda to supply the EU's AI gigafactories, a program covering up to seven sites and roughly EUR 30 billion, of which EUR 10 billion is public and approximately EUR 1 billion from Brussels is actually committed.

3

Today's Top Stories

© 2026 TheOutpost.AI All rights reserved