AI chip design

Nvidia has announced plans for a watchdog chip designed to sit alongside AI agents and monitor their behavior, according to a CNBC report. The idea is to give enterprises a hardware-level safety mechanism to supervise autonomous AI systems, flagging or halting actions that go out of bounds. The proposal is drawing attention and debate among developers about how AI safety should be enforced at the chip level.

The EU Chips Joint Undertaking has opened two calls for projects developing AI hardware, as part of Europe's push to strengthen its semiconductor capabilities. The move aims to fund research and development of chips designed for artificial intelligence workloads, supporting the EU's broader strategy to boost domestic chip production and reduce reliance on foreign suppliers. The initiative falls under European efforts to back next-generation computing technologies.

General Compute has deployed Cerebras' wafer-scale chips to speed up AI coding workloads, according to company announcements. The move pairs General Compute's infrastructure with Cerebras' large-format processors, which are designed to deliver faster inference than conventional GPUs. Details on deal size, rollout scale and customers have not been disclosed, but the deployment underscores growing demand for specialised hardware serving AI development tools.

A new science news release highlights work combining AI-driven chip design with III-V semiconductor pilot lines for electronics manufacturing. The initiative points to efforts to speed up development of compound-semiconductor devices and bring them closer to industrial production using artificial intelligence in the design process. Details on the organisations, funding and timelines involved are not provided in the available information.

Moore's Lab AI is the focus of a Semiconductor Engineering report examining how agentic AI can be applied across chip design workflows. The piece looks at where autonomous AI agents could take on tasks in semiconductor development, from design exploration to verification, and what it means for engineering teams. The discussion reflects broader industry interest in AI-driven automation in chipmaking.