Qualcomm just landed the kind of deal that makes Wall Street sit up and pay attention. The chipmaker’s shares jumped roughly 7-10% after Amazon announced a sweeping, multi-generational partnership to develop custom AI inference chips for AWS data centers.
The stock was trading near $180-$185 per share on the news.
What the deal actually involves
The partnership, announced on September 8, covers multiple generations of custom silicon specifically designed for AI inference workloads inside Amazon’s cloud infrastructure. Beyond the silicon itself, the collaboration includes advanced optical connectivity solutions capable of hitting speeds up to 1.6 terabits per second and beyond. Qualcomm is leveraging its own SerDes and optical DSP technologies to make that happen.
The deal also deepens Qualcomm’s existing relationship with Amazon on the software side. Qualcomm is increasing its reliance on AWS AI tools like Amazon Bedrock for its electronic design automation tasks.
Then there’s the equity component. Amazon received a warrant granting it the option to purchase up to 25 million Qualcomm shares at an exercise price of $161.26 per share. That warrant doesn’t expire until September 3, 2036, and at current trading levels, it represents roughly $4 billion in value.
Amazon gets an initial tranche of 3.75 million shares, but the rest is tied to commercial commitments that could total up to $60 billion.
Why this matters for Qualcomm’s identity
This Amazon deal is the clearest signal yet that Qualcomm’s diversification strategy is working. The company has now secured partnerships with three of the biggest hyperscalers in the world, following similar arrangements with Meta and Microsoft.
Qualcomm has raised its fiscal year 2029 targets to $15 billion in data-center revenue and $40 billion in total non-handset chip revenue.
Competitive landscape and what to watch
Qualcomm is entering a data-center AI market that already has formidable incumbents. Nvidia dominates AI training and inference GPUs. AMD has been gaining ground with its Instinct accelerators. And Amazon itself already designs custom AI chips through its Trainium and Inferentia lines via subsidiary Annapurna Labs.
The emphasis on inference rather than training is notable. Inference, the process of running trained AI models to generate outputs, is where the volume of compute demand is heading as AI applications scale to billions of users.
For investors, the warrant structure creates an interesting dynamic. Amazon’s incentive to vest the full 25 million shares means it has a financial reason to deepen its Qualcomm spending over the next decade. But the flip side is dilution risk: 25 million shares at $161.26 isn’t trivial.