なぜNvidiaの128GB「スーパーチップ」ノートPCはMacのAI支配を終わらせる可能性があるのか
nvidiaはwindowsノートPC向けに最大128gbのユニファイドメモリを備えるrtx spark「スーパーチップ」を発表し、appleのmacbook proに匹敵するローカルllm推論の実現を目指している。
nvidia announced the rtx spark platform at computex, combining a new arm-based n1 cpu, rtx graphics, and up to 128gb of unified memory to create 'superchips' designed for running large language models locally. unlike microsoft's previous copilot+ pcs, these systems aim to offer the memory capacity and gpu power previously exclusive to apple's macbook pro, specifically targeting developers and researchers needing local inference capabilities. nvidia plans to leverage its existing cuda software stack to ensure developers can access these hardware capabilities, potentially shifting the balance of power in local ai development back toward windows. this creates a direct competitor to apple's hardware, which has recently seen high demand due to its suitability for local ai tasks.