Qualcomm is adapting its High Bandwidth Compute architecture to bring data-center AI capabilities into smartphones and other devices, enabling on-device AI processing across mobile, PC, and automotive platforms.

The move reflects a broader shift away from cloud-reliant AI, leveraging Qualcomm’s chip expertise to expand into data centers and cars while unifying the AI ecosystem across devices.

The company is pursuing a common architecture to accelerate development and foster seamless AI capabilities from smartphones to PCs and vehicles.

If successful, future devices could run more advanced AI features, offer faster assistants, and improve battery life despite rising AI demands.

The on-device approach could let smartphones run larger AI models locally, cut cloud dependence, speed up responses, boost privacy, and support continuous or always-on AI without draining the battery.

The new architecture stacks memory and processing components vertically to shorten data travel within the chip, boosting speed and efficiency for AI workloads.

Qualcomm plans a phased rollout: initial deployment in data centers next year, commercial deployment by 2028, and eventual adoption in smartphones and other devices.