Edition Summary (20 Pages)

Google introduced Gemini 4 Carbon, emphasising cybersecurity and extended agent tasks while initially restricting access to selected defenders. Competition for AI infrastructure widened as Oracle prepared to offer Google TPUs, reports suggested OpenAI could purchase them, and chip demand lifted memory suppliers. Anthropic’s IPO filing drew scrutiny of its valuation, substantial compute commitments and potentially uninsurable AI risks, while Microsoft’s Azure growth and Oracle’s cash-flow forecasts sharpened attention on returns from infrastructure spending. Enterprise vendors pushed agent deployment towards stronger oversight: ServiceNow expanded ties with OpenAI and Google Cloud and launched a governance alliance, IBM introduced agent identity controls, and Salesforce added evaluation and cost safeguards. Lawsuits and research disputes reinforced questions over AI safety and reliability.