Two Years After Copilot+, Nvidia Is Shoving AI Into PCs Again - Gizmodo

Two years after the launch of Copilot+, NVIDIA and Microsoft are once again pushing AI integration into PCs with the introduction of ARM-based RTX Spark PCs. While Qualcomm had previously laid groundw... Read original →

NVIDIA’s RTX Spark push into ARM-based PCs is a tactical move to capture surging AI agent compute demand. Technically, ARM lacks high-performance GPU IP at 3nm EUV nodes, forcing local LLM inference to rely on the cloud—undermining Copilot+’s 'on-device intelligence' promise. Compliance-wise, U.S. export controls have driven Microsoft and Qualcomm to shift AI PC assembly to Vietnam and Taiwan, China, inflating BOM costs by over 15%. Strategically, Apple’s M-series chips—with unified memory and vertical integration—could lure enterprise users away from Windows, while ODMs like Asus face inventory risks from ARM ecosystem fragmentation. Over the next 18 months, unless Project Solara solves AI agents’ opaque resource consumption, the 'agentic lifestyle' will ring hollow. The real battleground is OS-level control over heterogeneous NPU+GPU scheduling—Microsoft is mimicking iOS’s walled garden but underestimating PC users’ entrenched expectation of computational sovereignty.