As of November 9, 2026, the AI industry’s center of gravity has shifted from raw model capability toward rules, money, and infrastructure.
A shared crawl signal
This week’s cleanest signal came from the web itself. Google, Microsoft, OpenAI and Anthropic said they will honor a single robots.txt line—”__usage_ai-training”—as a cross-company opt-out from AI training. The catch: it governs how content may be used, not whether bots can fetch it; it leaves already-scraped data untouched; and no major player fully honors it yet. Anthropic has begun, OpenAI promises “in the coming months,” Microsoft starts February 1, 2027, and Google targets 2027. The IETF has not endorsed it, and the draft lapses in May 2027 unless revised. It is progress, but not yet a standard.
An IPO reset, a $1.25T close
The money moved too. Anthropic’s IPO valuation expectations slipped to $1.2–1.3 trillion, down from $1.4–1.5 trillion earlier and the $2 trillion floated in October, after an AI-market rout. OpenAI closed the first $11 billion tranche of a round valuing it at $1.25 trillion—short of the $30 billion it originally sought—with President Greg Brockman hinting an IPO could come as early as 2027.
Gemini 4 Carbon tops the index
On capability, Google unveiled Gemini 4 Carbon, a gated cybersecurity model it says leads security benchmarks and tops the Artificial Analysis Intelligence Index v5.0 at 59, ahead of Claude Opus 5.5 (58) and GPT-6 Astra (57). It arrives with a 1M-token context window and introductory pricing of $2/$10 per million input/output tokens. Separately, Anthropic detailed “dreaming,” a technique that lets Claude replay session logs to consolidate memory—one agent wrote 413 memory files across 225 sessions.
The through-line: AI’s frontier is being negotiated in standards bodies, cap tables and data centers as much as in labs.
Sources: TechRadar · AIBriefs · Tilt · LLM Daily
Leave a Reply