chips

AI Index — Hardware (Stanford HAI)

The hardware chapter of Stanford's annual AI Index report: data on AI chip performance, costs, and who controls the computing power behind modern AI.

Why I recommend it: Free to download. It's the institute's own synthesis, and some compute and investment figures come from data supplied by the companies being measured — the broad picture is reliable, the fine print less so.

Google DeepMind Engineer Quits, Saying His Chip Work Could Speed Up AI

Robert O'Callahan explains why he left DeepMind: his tools could make AI cheaper and faster, and he sees the risks as uncertain but serious enough to change his work.

Why I recommend it: Free newsletter post. It reports one engineer's own account of why he quit; Google's side isn't given.

Tom's Hardware

Long-running hardware publication covering chips, GPUs, AI infrastructure and the machines behind it, free to read.

From the site: Tom

Why I recommend it: Useful if you're heading toward hardware, data centers or AI infrastructure and need to know what people actually run.