Qualcomm Secures Major Amazon Partnership to Develop Custom AI Data-Center Silicon

Qualcomm is targeting $15 billion in data-center sales in fiscal 2029 and has outlined a broader roadmap that includes an AI chip and technology for linking multiple chips together.
The Dragonfly C1000 is designed for agentic AI and is intended to deliver data-center computing performance with lower power consumption; Meta is expected to begin using the processor when production starts in 2028.
Amazon’s annual capital spending on AI infrastructure is described as reaching into the hundreds of billions of dollars, underscoring the scale of the potential market for Qualcomm’s data-center products.
Qualcomm’s recent financial and insider-trading data offered caution alongside the market’s positive reaction: the company reported $9.9 billion in third-quarter 2026 revenue, down 4.03% year over year, while insiders recorded 70 sales and no purchases over the prior six months.
Qualcomm and Amazon Web Services announced a major partnership to build custom AI chips for data centers, marking a significant win for Qualcomm's push beyond smartphones. IT Brief reported the deal involves multiple generations of processors designed for AI inference — the technology that powers chatbots and AI assistants after they've been trained. Qualcomm's stock jumped roughly 7% to 10% after the announcement, as investors saw it as proof the company's data-center strategy is working.
The partnership expands Qualcomm's customer base beyond Meta and signals Amazon's confidence in the company's chip-design abilities. The Verge noted the companies will also develop high-performance optical connectivity that can handle up to 1.6 terabits per second, allowing multiple chips in massive data centers to communicate faster. This is crucial as AI systems grow larger and more complex.
At the heart of the deal is Qualcomm's Dragonfly C1000 processor, designed specifically for running AI inference workloads with lower power consumption than competitors. Hot Hardware reported the companies will co-develop multiple generations of custom chips optimized for Amazon's hyperscale data centers. Meta is expected to begin using the Dragonfly processor when production starts in 2028, giving Qualcomm two major hyperscalers as customers.
Qualcomm targets $15 billion in data-center sales by fiscal 2029, a dramatic shift from its traditional smartphone business. The Dragonfly is designed for agentic AI — systems that can think and act independently — making it a key weapon in Qualcomm's effort to capture a slice of the enormous AI infrastructure market. Power efficiency matters because data centers consume massive amounts of electricity, and every watt saved cuts costs.
Amazon spends hundreds of billions of dollars annually on AI infrastructure, according to the partnership announcement. This massive budget gives Qualcomm a clear picture of where the market is heading and provides guaranteed demand for its chips. AI Authority described the collaboration as building "next-generation AI data center infrastructure," underscoring how critical this shift is for Qualcomm's future revenue growth.
The partnership also includes using Amazon's Bedrock service — a tool for building AI applications — to design and test the chips. This means Qualcomm benefits from AWS infrastructure while developing products specifically tailored to Amazon's needs. It's a win-win that locks in a long-term customer relationship.
While the stock market cheered the Amazon deal, warning signs lurk beneath the surface. Qualcomm reported $9.9 billion in third-quarter 2026 revenue, down 4.03% year over year, reflecting weakness in its core smartphone business. The smartphone market is mature and saturated, forcing Qualcomm to find growth elsewhere — hence the aggressive push into data centers.
Insider selling added to the caution. Over the prior six months, Qualcomm insiders recorded 70 stock sales and zero purchases, suggesting company executives may doubt near-term prospects. The Amazon deal is impressive on paper, but Qualcomm must prove it can execute and deliver profits from this new market before investors fully believe in the turnaround.
Publishers
109
Articles
310
Reach
419