Qualcomm closed roughly 12% higher on Friday and is up about 75% over the past month, trading at a record, as Wall Street reprices the company around three converging bets: data-center inference silicon, automotive design wins, and edge AI across phones, glasses, cars, and robots.
What moved the stock
The immediate catalyst was a May 21 expansion of Qualcomm's partnership with Stellantis. The automaker committed to deploying Qualcomm's Snapdragon Digital Chassis system-on-chips — covering cockpit, connectivity, and advanced driver-assistance — across its next-generation vehicle architectures. The deal includes the Snapdragon Ride Pilot platform for Level 2+ hands-free automated driving, integrated with Stellantis' STLA Brain software stack, standardizing Qualcomm compute across the company's brands and locking in a multi-year automotive revenue base beyond handsets.
The rally is broader than one contract. Investors are, in CNBC's framing, "waking up" to Qualcomm's pivot from a smartphone-modem story to a diversified AI compute supplier.
The data center play
The most consequential move for infra buyers is Qualcomm's re-entry into the data center. Its AI200 (shipping 2026) and AI250 (2027) are inference accelerators built on the Hexagon NPU, targeting Nvidia's roughly 80% share on total cost of ownership rather than raw training throughput. The AI200 supports 768GB of LPDDR per card; the AI250 adds a near-memory computing design Qualcomm says delivers about 10x effective memory bandwidth. The parts ship as rack-scale, direct-liquid-cooled systems at a 160 kW power envelope.
Qualcomm has also said it is collaborating with Nvidia on AI cluster solutions, including NVLink integration — positioning its accelerators inside, not only against, Nvidia-centric racks. Saudi Arabia's HUMAIN is the anchor customer, targeting 200 MW of AI200/AI250 deployments starting in 2026.
OpenAI and the edge
Qualcomm is also reportedly working with OpenAI on a chip for an upcoming device, reinforcing the thesis that inference is moving to the edge — phones, wearables, and dedicated AI hardware — where Qualcomm's power-efficiency lead matters more than data-center FLOPS. Its reported $2.4 billion acquisition of Alphawave adds the high-speed interconnect IP needed to compete at rack scale.
What it means for builders
For infra teams, the signal worth tracking is Qualcomm becoming a credible third inference vendor alongside Nvidia and AMD — more than the stock move itself. A liquid-cooled, LPDDR-heavy, TCO-optimized rack changes the procurement math for inference-bound workloads, and NVLink compatibility lowers the switching cost. For teams shipping on-device AI, the Stellantis and OpenAI signals suggest Snapdragon will be a default target for production inference at the edge through 2027. The open question is software maturity: whether Qualcomm's stack can close the gap with CUDA's tooling fast enough to convert design wins into deployed capacity.
