Intel and Nvidia Roll Out New AI Chip Capabilities for PCs

Intel’s dormant Neural Processing Unit (NPU) has been reactivated by developers using a large language model to rewrite its firmware, delivering a 3.2‑fold speed increase for AI inference on laptops and edge devices. The effort unlocks the chip’s full 128 TOPS potential, promising lower power consumption and reduced thermal throttling, though experts warn the unofficial firmware could introduce security risks.

At Computex in Taipei, Nvidia announced its RTX Spark processor, a hybrid super‑chip aimed at the $200 billion CPU market. The 1‑peta‑flop chip can run large language models locally and safely sandbox AI agents, with partner OEMs such as Dell, HP, Lenovo and Microsoft set to ship PCs equipped with the new processor in the autumn. Nvidia positions RTX Spark as a step toward PCs that act as autonomous AI assistants.

Both announcements signal a shift toward integrating powerful AI acceleration directly into consumer PCs, challenging traditional GPU‑centric approaches and expanding the role of AI hardware in everyday computing.