| Mobile Web

Search results for KV Cache

Industry

Snapdragon Summit 2026: Smartphones can run 30 billion-parameter AI models

On-device AI models, previously constrained by smartphone memory capacity, can grow to 30 billion parameters through data placement design, Qualcomm said. At Snapdragon Summit 2026, Qualcomm executive vice president Chris Patrick said more capable AI agents need access to larger models with stronger reasoning and richer context. Qualcomm unveiled the Snapdragon 8 Elite 6th Gen platform, expanding Hexagon NPU shared memory and adding an element accelerator supporting 32,000 tokens.

Industry

Qualcomm boosts Hexagon NPU memory 50 percent, targets AI agents

Qualcomm has redesigned its AI processor architecture to run agentic AI on smartphones. The company on Sept. 11 disclosed design details and performance figures for the Hexagon NPU to be used in its next premium mobile platform. It added a new accelerator for transformer workloads and increased shared NPU memory by up to 50 percent to reduce DDR access and latency. Qualcomm also outlined wider precision support and MoE model support.

AI & Enterprise

Lightbits launches KV cache engine to boost GPU performance

Lightbits Labs has launched its AI inference software engine, Infera, aimed at improving inference performance and cost efficiency. The company says Infera expands KV cache data beyond limited high-bandwidth memory attached to GPUs and targets large language model operations with long context windows or many concurrent sessions. It manages KV cache across GPU memory, DRAM and NVMe storage using predictive prefetching and supplies context in small blocks.