Qualcomm Launches AI Chips for Smartphones with 30B Model

Qualcomm introduces two AI‑focused smartphone chips, with the flagship supporting a 30‑billion‑parameter MoE model on‑device, enhancing privacy and speed.

Qualcomm unveiled two new AI‑focused smartphone chips, with its flagship processor capable of running a 30‑billion‑parameter mixture‑of‑experts (MoE) model locally.

The company said the ability to host a MoE model on a handset marks a step toward more sophisticated on‑device artificial‑intelligence tasks. By keeping inference inside the phone, developers can reduce latency and protect user data.

Why on‑device AI matters

Running large language models on a smartphone eliminates the need for constant cloud connectivity. Users benefit from faster response times for features such as real‑time translation, image enhancement and personalized assistants. The local execution also sidesteps bandwidth costs and regulatory concerns tied to data transmission.

Implications for smartphone makers

Manufacturers that adopt Qualcomm’s new chips can differentiate their products with richer AI experiences. The chips are positioned to support next‑generation applications that demand high‑capacity models without sacrificing battery life, although specific power‑efficiency figures were not disclosed.

Industry analysts note that the move aligns with a broader trend of embedding more compute power in mobile silicon. Competitors are also exploring on‑device AI, but Qualcomm’s claim of handling a 30‑billion‑parameter MoE model sets a clear benchmark.

Qualcomm’s roadmap

Qualcomm indicated that the two chips will enter the market later this year, targeting flagship and premium‑tier devices. The company did not provide exact model names or release dates, focusing instead on the functional milestone of local MoE execution.

By delivering a processor that can run a model of this scale, Qualcomm aims to enable developers to create applications previously limited to data‑center hardware. The announcement underscores the firm’s commitment to advancing mobile AI while keeping the user experience on the device.