The dominant story this week was consolidation and institutionalization: AI is no longer being piloted at the edges of organizations — it's being wired into the core.
Anthropic had the most consequential week of any lab. The company confidentially filed IPO paperwork with the SEC, revealing a revenue run rate of $47 billion — up from $10 billion just a year ago — and a pre-IPO valuation topping $965 billion, ahead of OpenAI. A listing is targeted for fall 2026. Simultaneously, Anthropic announced the expansion of Project Glasswing to approximately 150 new organizations across more than 15 countries, deploying Claude Mythos to hunt vulnerabilities in power, water, healthcare, and communications infrastructure. Partners including Apple, Nvidia, Microsoft, CrowdStrike, and Palo Alto Networks have already surfaced over 10,000 high- or critical-severity vulnerabilities, including zero-days across all major operating systems and browsers. It was a rare week where the same company announced both a Wall Street ambition and a tangible safety mission at scale.
Microsoft Build 2026 (June 2–3, San Francisco) provided the industry's backdrop for much of the week's activity. Microsoft introduced MAI-Thinking-1, a 35-billion-parameter reasoning model with a 256K context window entering private preview on Azure Foundry, alongside Autopilots — long-running autonomous agents anchored by Scout — and the IQ context layer for feeding agents real-time workplace data. The enterprise adoption numbers were equally striking: Infosys, TCS, and Wipro collectively crossed 300,000 Microsoft 365 Copilot seats in under six months, the clearest signal yet that large IT services firms are treating Copilot as standard tooling.
OpenAI expanded its distribution on two fronts: frontier models including GPT-5.5 and Codex are now available on Amazon Web Services via Bedrock, and the company launched OpenAI for Government with compliance-oriented terms for public-sector deployments. Codex also reached 5 million weekly users, with non-developers now accounting for 20% of usage and growing three times faster than developers — a sign that coding agents are finding audiences well beyond engineering teams.
On the secondary front, Google's Gemini AI Mode — now used by over one billion monthly users — is intensifying advertiser anxiety as it answers more queries without generating outbound traffic, accelerating the collapse of the web-search referral model. GitHub released the Copilot SDK as generally available, repositioning Copilot as an embeddable platform for third parties. Microsoft also open-sourced the Agent Control Specification, a runtime policy standard compatible with LangChain, AutoGen, CrewAI, and other major agent frameworks. China produced another AI unicorn: Vast, a 3D-modeling startup, reached a $1 billion valuation on the back of gaming-industry demand.
To watch next week: Anthropic's IPO process will likely dominate the news cycle as analysts begin digesting its financials and the broader question of AI company valuations comes into focus.