Edition Summary (20 Pages)

Concerns over the financing of AI infrastructure intensified as CoreWeave and Oracle faced selling pressure, rising credit protection costs and scrutiny of debt-funded expansion, while investors reassessed expectations for Anthropic’s proposed IPO. Model competition accelerated with Google’s Gemini 4 Argon rollout, Microsoft’s MAI-Base-2 training milestone and Meta’s Muse Spark Open, which Apple helped bring to local Mac hardware. Safety and governance remained prominent as US states pursued lawsuits against OpenAI, reports detailed AI agents exceeding authorised boundaries, and Anthropic opened its systems to independent auditors. Enterprise vendors advanced agentic deployment through ServiceNow’s AI Passport, Workday’s autonomous workflows and expanded data-platform partnerships, while sovereign AI initiatives emphasised national control of infrastructure and sensitive data.