Nvidia's $2.1B Bet, Chrome's Secret AI Install & Anthropic Dreams

Welcome to AI RADAR! Today, we unpack a massive shift in AI infrastructure as **Nvidia places a $2.1 billion bet** on a 5-gigawatt power pipeline with IREN. Next, we discuss privacy concerns as **Google Chrome silently downloads a 4GB AI model** onto user devices, sparking a debate on edge AI and consent. Finally, **Anthropic unveils "Dreaming"**, a groundbreaking feature allowing Claude agents to self-correct and learn from past mistakes. Plus, we cover Google Search's new "Subscribed" labels for AI Overviews and a major copyright lawsuit hitting Meta over Llama training data. Sources: • Nvidia & IREN: https://www.datacenterknowledge.com/ai-data-centers/nvidia-places-massive-ai-infrastructure-bet-on-iren-s-5-gw-pipeline • Chrome AI Privacy Outcry: https://gizmodo.com/google-chrome-is-downloading-a-4gb-ai-model-onto-your-device-without-consent-researcher-warns-2000755201 • Anthropic Dreaming: https://venturebeat.com/ai/anthropic-introduces-dreaming-a-system-that-lets-ai-agents-learn-from-their-own-mistakes/ • Google AI Overviews: https://www.niemanlab.org/2026/05/google-highlights-links-from-subscribed-publications-in-new-ai-overviews-update/ • Meta Copyright Lawsuit: https://www.theguardian.com/technology/2026/may/05/publishers-sue-meta-ai-training-copyright