Anthropic hits $965B valuation, files for IPO amid major model and rival releases

Anthropic released Claude Opus 4.8 with improved benchmark scores and introduced Dynamic Workflows, a tool designed to support long-running multi-agent tasks. The release included discussion of eval-awareness findings and themes around welfare and corrigibility drawn from the model's system card. Simultaneously, Anthropic announced that it had raised $65B in Series H funding at a $965B valuation and filed for an IPO, signaling its transition toward public-market status.

Microsoft unveiled Scout, an always-on AI personal assistant built on OpenClaw, alongside new in-house MAI models—including MAI Thinking 1—and a technique called "frontier tuning." The company emphasized enterprise security architecture and its capability to develop models from scratch, positioning these moves as a shift toward reduced dependence on external model providers. In a separate funding milestone, Cognition, an AI coding startup, raised $1B at a $25B pre-money valuation. Open-source competition intensified when Minimax-M3 debuted, claiming to match or exceed GPT-5.5 and Gemini 3.1 Pro on key benchmarks while requiring only 5–10% of the cost.

A JPMorgan analysis published during the same period argued that OpenAI needs a 26× revenue increase to justify its infrastructure spend, underscoring doubts about whether current business models can sustain the sector's capital intensity. Policy moves added another layer of complexity: Trump signed an executive order establishing a voluntary pre-release government testing framework for powerful AI models. The U.S. tightened controls on Nvidia AI chip exports, and China began requiring private-sector AI experts to secure approval before international travel—measures framed as efforts to secure top-tier talent. OpenAI launched Rosalind Biodefense and offered federal agencies early access to its life-sciences model. Security incidents also surfaced: Meta AI was exploited by hackers who manipulated the system to gain access to high-profile Instagram accounts simply by asking.

The episode captures a pivotal moment in AI commercialization: Anthropic's $965B valuation and IPO filing represent the sector's confidence in scaling, yet a parallel JPMorgan analysis reveals a structural tension. That analysis found OpenAI requires a 26× revenue increase to justify its infrastructure spend—a sobering metric that suggests the sector's spending trajectory may outpace revenue growth for years. Against this backdrop, Minimax-M3's claim to deliver frontier-model performance at 5–10% of the cost signals that open-source alternatives are closing the efficiency gap, potentially disrupting the pricing power of commercial leaders.

Microsoft's announcements of Scout and new MAI models (including MAI Thinking 1) position the company to compete directly with OpenAI and Anthropic while retaining in-house model control and enterprise security architecture—a strategic shift toward vertical integration. Cognition's $1B funding at a $25B valuation reflects investor appetite for AI-native software startups, distinct from the foundational-model segment. On the policy front, Trump's voluntary pre-release testing framework and new U.S. export controls on Nvidia chips are beginning to shape which models and hardware configurations enterprises can adopt, potentially fragmenting the global AI market.

No comments yet. Be the first to share your thoughts!