Amazon이 Qualcomm AI 칩에 $60 billion을 투자한다.

Qualcomm and Amazon signed a multi-generation agreement on September 8, 2026, covering customized silicon for AWS AI infrastructure, with the first phase aimed at inference. The two companies will develop high-performance optical connectivity extending to 1.6 Tb/s and beyond, using SerDes and optical DSP technology from Qualcomm's Alphawave acquisition. Qualcomm also plans to deepen its use of AWS infrastructure, including Amazon Bedrock, for electronic design automation workloads to target shorter chip design cycles.

The regulatory filing reveals the true magnitude of this relationship. Amazon received a warrant to purchase up to 25 million Qualcomm common shares at $161.26 each, worth roughly $4 billion at issuance. The warrants vest in stages tied to commercial agreements, binding purchase orders, and Amazon's actual purchases of Qualcomm server chips, technology, systems, and manufacturing services. The vesting schedule tops out at a maximum threshold of $60 billion in payments from Amazon over the warrant's decade-long life through September 2036. Qualcomm already vested warrants covering 3.75 million shares at issuance against initial purchase commitments.

The $60 billion figure sets the ceiling of the vesting ladder—the maximum purchase-based threshold for full warrant vesting—giving both sides genuine incentive to grow the relationship over time. It carries no purchase commitment, guaranteed contract value, or committed backlog. The structure pays Amazon in Qualcomm equity for buying Qualcomm silicon, aligning both parties toward volume growth and giving Qualcomm a major reference customer at hyperscale. Marvell structured a similar deal with Google in August, granting a warrant for approximately 59 million shares worth up to $12.2 billion.

The scope extends well beyond an accelerator. Initial coverage framed this as a custom AI chip deal, but the filing describes server chip products, technology, complete systems, manufacturing services, and the optical connectivity that ties AI infrastructure together. That maps onto the Dragonfly portfolio Qualcomm outlined in June 2026, spanning custom silicon, CPUs, AI inference accelerators, and connectivity products, built on strength in energy-efficient processing, digital signal processing, high-speed SerDes, and system-level semiconductor integration. Meta signed for the C1000 in June for its next-generation server fleet, and Microsoft backs the data center push. Amazon becomes the third hyperscaler on the list, and the one paying with equity upside.

Silicon teams should focus on the inference framing. While training large foundation models captures most industry attention, the cumulative arithmetic of inference—repeated execution of trained models against billions of user and application requests—is on track to become the larger long-term computational workload. Inference rewards narrow, specialized architectures. A design tuned for particular models, fixed numeric format, specific memory hierarchy, and defined thermal envelope can outperform general-purpose parts on tokens per dollar. GPUs deliver broad programmability and mature software environments. Custom accelerators deliver efficiency within a narrower problem definition. At Amazon's fleet scale, even modest gains of a few percentage points per watt translate into substantial reductions in infrastructure and operating costs.

Connectivity carries equal weight in this agreement. The optical component demonstrates why Qualcomm paid $2.4 billion for Alphawave Semi and closed the deal in December 2025. Alphawave brought high-speed SerDes, optical DSPs, chiplets, and custom silicon expertise—technology Qualcomm explicitly positioned as accelerating its data center expansion by complementing existing Oryon CPU and Hexagon NPU technology with the high-speed connectivity needed for larger systems. Former Alphawave CEO Tony Pialis now runs Qualcomm's data center chip business.

AI system performance increasingly depends on moving data between accelerators, memory, servers, racks, and buildings, not solely on raw processor speed. Accelerators wait on memory, neighboring dies, top-of-rack switches, and inter-building links. A vendor supplying the accelerator, CPU, SerDes, and optical DSP can co-design across these boundaries. Qualcomm's combination of existing compute and AI technology with Alphawave's connectivity portfolio positions it as a provider of multiple technologies required to build and connect complete AI infrastructure—separating it from suppliers offering compute alone or networking alone.

The agreement does not indicate Amazon is replacing Nvidia GPUs or abandoning its own internally developed processors. AWS already operates one of the industry's broadest computing architecture collections, spanning Nvidia and AMD accelerators, its own Trainium and Inferentia AI processors, Graviton CPUs, and Nitro infrastructure silicon. Amazon's custom chip business passed a $25 billion annualized run rate by the end of the June quarter. Adding Qualcomm reinforces Amazon's ability to match the best architecture to each workload, maintaining architectural and supplier flexibility and negotiating leverage. Qualcomm and Amazon have disclosed nothing about how the joint products relate to Trainium and Inferentia roadmaps.

Qualcomm previously entered the server processor market through its Arm-based Centriq family but withdrew before reaching commercial scale. Today's market operates differently: cloud providers now rank among the world's largest semiconductor customers, increasingly willing to deploy custom silicon to control cost, power consumption, supply, and product differentiation. Generative AI has created demand for new compute and networking architectures at unprecedented scale. Qualcomm also enters this market with a far broader asset base than during the Centriq era, including Oryon CPU cores, mature AI acceleration technology, an expanding AI software platform, and the SerDes, optical DSP, chiplet, and custom silicon capabilities Alphawave added.

"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, Qualcomm's president and CEO. Prasad Kalyanaraman, AWS vice president, frames the deal from Amazon's perspective around continuity: "By working together on customized silicon and advanced connectivity, we're delivering more performant, efficient, and cost-effective infrastructure for our customers."

For independent software vendors, Qualcomm's inference stack will need to absorb the model formats and serving frameworks AWS customers already run, with porting costs landing on the software layer. Questions to ask include compiler maturity, quantization support, kernel coverage for attention variants, and the migration path from CUDA-shaped code.

For CIOs and IT decision-makers, this presents a supply question with a 2028 horizon. Neither the C1000 nor the AI300 ships before then, so procurement teams have two years to model the effect on AWS instance pricing and reserved capacity plans. A third viable inference architecture inside AWS pushes down per-token cost while also fragmenting the tooling surface platform teams must support.

The warrant structure tells the real story here. A $60 billion maximum threshold, even as an aspirational ceiling, signals Qualcomm's genuine confidence in sustained Amazon business. The optical connectivity angle deserves equal weight: Qualcomm's Alphawave acquisition now reads as a deliberate bet that AI infrastructure competition runs as much through data movement as raw compute, with Amazon validating that thesis through substantial commitment.

---

**Note:** The source text appears to end mid-sentence ("Amazon validating that thesis with rea"). I've completed it with the most logical conclusion based on the article's arc. If you have the complete ending, I can refine it further.