OpenAI and Anthropic Race to Wall Street
OpenAI filed a confidential S-1 registration statement with the SEC on Sunday, June 8, confirming its intention to go public at a staggering $852 billion post-money valuation. The filing comes just one week after rival Anthropic submitted its own confidential IPO paperwork on June 1, setting the stage for what could be the most significant tech IPO race since the dot-com era.
OpenAI, backed by 900 million weekly active ChatGPT users and roughly $2 billion in monthly revenue, tapped Goldman Sachs and Morgan Stanley as underwriters. However, the company faces significant headwinds: it has missed recent targets for new users and revenue, projects spending $122 billion on compute in 2028, and doesn't expect positive cash flow for at least four more years.
Anthropic, meanwhile, appears to be in a stronger financial position. Following a massive $65 billion Series H that valued the company at $965 billion, Anthropic claims to be nearing its first quarterly profit. On secondary markets, the company recently surged past the $1 trillion mark, surpassing OpenAI's valuation — a remarkable reversal for a company founded by former OpenAI executives just a few years ago. With SpaceX also eyeing a $1.75 trillion listing, 2026 is shaping up to be a historic year for tech IPOs.
Anthropic Warns AI May Soon “Build Itself” — Calls for Coordinated Global Pause
Days after filing for its IPO, Anthropic published one of the most consequential AI safety documents in recent memory. “When AI builds itself,” released on June 4, argues that AI systems are approaching the threshold of recursive self-improvement — the point where models can meaningfully improve their own successors with minimal human intervention.
The numbers are striking: over 80% of code merged into Anthropic's production codebase is now authored by Claude, up from single digits before February 2025. The company's engineers ship roughly eight times more code per day than they did in 2024. Claude's success rate on open-ended research problems reached 76% in May 2026, climbing 50 percentage points in just six months. Perhaps most remarkably, Claude outperformed human researchers in choosing research directions 64% of the time.
Anthropic outlines three possible futures: capability trends stalling (assessed as unlikely), compounding efficiency gains where 100-person teams accomplish what once required 10,000+ people, or full recursive self-improvement where progress becomes limited only by compute availability. The company's central recommendation is a globally coordinated mechanism to pause or slow frontier AI development — but only if verifiable conditions ensure all major labs participate simultaneously.
The proposal has drawn both praise and skepticism. Critics have noted the timing: Anthropic filed for its IPO just days before publishing the call for a potential pause, raising questions about whether the report is a genuine safety warning or strategic positioning ahead of its public market debut.
Trump Signs AI Executive Order: Voluntary 30-Day Review for Frontier Models
On June 2, President Trump signed the executive order “Promoting Advanced Artificial Intelligence Innovation and Security,” establishing a voluntary framework for AI companies to submit their most powerful models for government review up to 30 days before public release.
The order directs federal agencies to develop benchmarks for assessing AI models' cybersecurity capabilities, create an “AI cybersecurity clearinghouse” for sharing vulnerability information, and strengthen government security defenses. Notably, it explicitly states that nothing in the order authorizes mandatory licensing, pre-clearance, or permitting for AI development — a clear nod to the administration's innovation-first approach.
The 30-day review window represents a compromise: an earlier draft reportedly proposed 90 days, but pushback from industry allies led to the shorter timeline. Meanwhile, at the state level, Colorado's comprehensive AI Act is set to take effect on June 30, requiring security risk management programs, impact assessments, and measures to prevent algorithmic discrimination for high-risk AI systems — highlighting the growing patchwork of AI regulation across the United States.
The AI Coding Wars Heat Up: Microsoft and Google Challenge Claude's Dominance
The AI-assisted coding market — arguably the most commercially important AI application today — saw major moves in early June. Anthropic's Claude Code has emerged as the dominant force in the space, but Microsoft and Google are mounting aggressive counterattacks.
At its Build developer conference, Microsoft unveiled MAI-Code-1-Flash, a 5-billion-parameter coding model now rolling out across Visual Studio Code and GitHub Copilot. The model is part of a broader strategy: Microsoft announced seven homegrown AI models at Build, signaling a push toward “long-term self-sufficiency” and reduced reliance on partner OpenAI.
Google's challenge is more measured. CEO Sundar Pichai candidly acknowledged that “when it comes to agentic coding with tool use, instruction following, and long-horizon tasks, I think we are a bit behind.” The company is competing on price, offering an AI developer subscription tier at $100 per month, and is expected to release Gemini 3.5 Pro this month — which could shake up the competitive landscape if it delivers on the coding capabilities previewed at Google I/O.
The stakes are enormous. AI-assisted coding is one of the few AI applications with clear, measurable ROI, but profitability remains elusive: recent analyses suggest that LLM-assisted coding may cost providers more than $1,000 for every $100 users pay, with subscriptions heavily subsidized by venture funding and corporate balance sheets.