AI infrastructure

AI Headline Hub’s reporting on AI infrastructure, including chips, training computers, data centers, and agent runtimes. Each story links its primary source.

This page groups chips, training computers, data-center plans, and the machines vendors say agents run on.

The newest file is Google’s Suncatcher prototype, which the company says ran a Gemma 4 model on a TPU in orbit. Before it: Elon Musk’s statement, via Reuters, that SpaceX closed $40 billion in debt for GPUs, Meta’s stock sale for AI spending, Anwar’s account of a SoftBank plan for Malaysia, and Cadence and Nvidia’s Maestro models for chip design. Lula’s pledge to attract data centers is filed here too. Earlier files include Commerce’s letters of intent for power-chip research, SageMaker AI training sessions, and Google Cloud’s Agent Compute Engine preview. The poll on data centers is the public-opinion file in this set.

Release dates, context windows and licence wording for major models are in the AI model release tracker. The same rows are in a CSV.

Stories

Explainers

Other topics