Anthropic overtakes OpenAI at $30B run rate

Bloomberg

Anthropic's revenue run rate has skyrocketed to $30 billion from $9 billion in just months, driven by surging enterprise demand for Claude, with 1,000+ customers spending over $1 million annually and partnerships with Broadcom and Google supporting its growth.

TechCrunch

Anthropic launched Mythos, its most powerful AI model yet, exclusively for defensive cybersecurity work with major tech partners, claiming it has already discovered thousands of critical zero-day vulnerabilities. The limited release reflects concerns about weaponization while aiming to strengthen critical infrastructure security across the industry.

TechCrunch

OpenAI has proposed a comprehensive economic policy framework for the AI age, including public wealth funds, robot taxes, shifting tax burden from labor to capital, and subsidized four-day workweeks, positioning itself in the debate over how to distribute AI-driven prosperity amid growing concerns about job displacement and wealth concentration.

Meta AI

Meta unveiled Muse Spark, a multimodal reasoning model with multi-agent orchestration that requires 10x less compute than their previous models, marking their first step toward 'personal superintelligence' with applications in health, visual reasoning, and agentic tasks. The model is available at meta.ai with a private API preview for select users.

The Register

Anthropic has restricted Claude subscription use with third-party AI agent tools like OpenClaw due to capacity constraints and system strain, forcing heavy users to pay for extra usage or switch to more expensive API access. The enforcement reflects broader infrastructure challenges as AI demand outpaces datacenter availability.