AI Weekly

Welcome to AI Weekly! ChatGPT hits one billion monthly users and Anthropic quietly files for an IPO in the same week, xAI lands its first major federal government contract, and a leaked source map suggests Anthropic's next model is closer than anyone announced. Let's process the intelligence that matters.

🔥 THE BIG THREE

1. ChatGPT Hits a Billion Users. Anthropic Files for an IPO. Same Week.

ChatGPT crossed one billion monthly active users in May 2026, making it the fastest app in history to hit that scale, ahead of TikTok, Instagram, and YouTube, in roughly two and a half years since launch. The milestone landed the same week Anthropic made its own move, confidentially filing a draft S-1 registration statement with the SEC on June 1, 2026, with a revenue run-rate hitting approximately $47 billion in May 2026, up roughly 5x from the prior year. Both OpenAI and Anthropic are now on the IPO track simultaneously, and the race to define "the AI company" for public markets is underway. Stark InsiderIEEE Spectrum

Why This Matters: The billion-user number sounds clean, but more data is underneath it. Claude has 56 million monthly active users growing at 640 percent year-over-year while ChatGPT grew 62 percent, and US users who installed Claude in Q1 2026 spent 5 percent less time on ChatGPT just one month later. That substitution is small today and enormous at scale. Meanwhile, Anthropic's Claude models account for roughly 40 percent of enterprise LLM spending compared to OpenAI's 27 percent. ChatGPT owns consumer. Claude is winning enterprise. Both are heading to public markets to make their case to institutional investors at the same time. Stark InsiderStark Insider

What's Next: OpenAI is projected to file its own S-1 within weeks. Whichever company goes first shapes the valuation narrative for the other. Watch the prospectus language around enterprise market share and profitability timelines, because those two numbers define how Wall Street values the AI sector over the next several years.

2. xAI Lands a Sweeping Federal Government Contract

The US government awarded xAI a major federal AI contract this week, with Grok for Government rolling out across agencies. Combined with the launch of Grok Build for developers, this week represents xAI's most significant product push since the company was created. Since SpaceX's acquisition of xAI earlier this year, the company has transformed from a model lab into what observers are calling a full enterprise AI stack with government, developer, and consumer distribution channels simultaneously. OpenAI responded by granting the European Union access to GPT-5.5-Cyber, a cybersecurity-focused variant rolling out to vetted EU teams, governments, and institutions. Deadline

Why This Matters: When xAI goes from consumer chatbot to federal government contractor in one week, it signals that the government AI procurement race is no longer a two-horse OpenAI versus Anthropic competition. SpaceX's infrastructure relationships and Elon Musk's political access have opened a lane that the AI labs cannot replicate. Anthropic expanded Project Glasswing on June 2 to cover power, water, healthcare, communications, and hardware sectors, but has not yet granted EU access to Claude Mythos, giving OpenAI a strategic differentiator in the race for European government contracts. CNBC

What's Next: Government AI procurement is becoming the most consequential revenue category in the industry. The winners of federal and allied-nation contracts will have capital advantages that compound for a decade. Watch for Anthropic's EU move and whether OpenAI's GPT-5.5-Cyber creates a meaningful wedge in European institutional relationships.

3. A Leaked Source Map Points to Claude Sonnet 4.8 Coming Soon

Evidence for an intermediate Claude Sonnet 4.8 has been circulating since a source map was accidentally shipped with the Anthropic Claude Code npm package in March 2026. Inside that map, a security filter list contained three strings: sonnet-4-8, opus-4-7, and mythos. Opus 4.7 subsequently shipped on April 16, exactly as the leak implied, giving the other two strings real credibility. A mid-June Sonnet release is widely anticipated in developer communities. CNBC

Why This Matters: Anthropic's release cadence has never skipped a minor version, which means going from Sonnet 4.6 directly to Sonnet 4.8 would be unprecedented. The more likely read is that a Sonnet 4.7 shipped quietly or the numbering reflects an internal milestone. Either way, benchmark leaks attributed to OpenAI testers are also circulating, consistent with a June 2026 release window for GPT-5.6, with expected improvements in reasoning accuracy for multi-step agentic workflows. June is shaping up to be the most model-dense month of the year when both companies are trying to impress public markets. CNBC

What's Next: If Sonnet 4.8 ships before the Anthropic S-1 goes public, it becomes part of the IPO story. A frontier model release timed to an IPO filing is not accidental. Watch the Anthropic developer channels in the second and third week of June closely.

📊 WHAT ELSE WE'RE WATCHING

GitHub Copilot Goes Token-Based: GitHub Copilot switched from flat subscription pricing to token-based billing on June 1, 2026 changing the economics of AI coding tools and raising questions about what "unlimited" AI access will mean going forward Stanford

SoftBank Bets $87.5 Billion on France: SoftBank announced at the Choose France summit that it will invest up to $87.5 billion to develop 5 gigawatts of AI data center capacity in France, the largest announced AI infrastructure investment in European history Stanford

Uber Robotaxis Hit Spain: Uber launched robotaxi service in Spain this week, expanding autonomous vehicle deployment beyond the US for the first time, at scale anyway

OpenAI Finalizes IPO Paperwork: OpenAI is finalizing its Wall Street IPO paperwork with SpaceX's IPO roadshow completing its first full week, pricing scheduled for June 11 under the ticker SPCX Deadline

🛠️ AI TOOL SPOTLIGHT

GPT-Rosalind for Life Sciences: OpenAI released an updated GPT-Rosalind model on June 4, purpose-built for life sciences research at enterprise scale, combining GPT-5.5's agentic capabilities with stronger model intelligence across drug discovery, genomics, proteomics, and spatial transcriptomics. On GeneBench, it uses 31 percent fewer tokens than GPT-5.5 while achieving higher accuracy, which could be a big deal. Deadline

Why it matters: Vertical AI built for scientific domains is where the next wave of real-world AI lands. A model handling drug discovery end-to-end with fewer compute requirements changes the economics of pharmaceutical research. This could be the first of a scientific workflow replacement.

💭 CLOSING INSIGHT

This week put two numbers next to each other that tell the whole story: ChatGPT at one billion users, Anthropic at 640 percent annual growth. OpenAI has the distribution. Anthropic has the trajectory. Both are filing for IPOs at the same time. xAI just landed a federal government contract by running a different play entirely. And buried in a developer npm package, a leaked string suggests Anthropic's next model drops before the month is out.

The AI industry has entered its public markets era. Everything that happens from here gets priced by analysts who have never had to evaluate a company growing 5x per year with no clear path to profitability. This is new territory, and I would argue perhaps for everyone since it could be bigger than the dot com bubble.

Poll: When Anthropic and OpenAI go public, which company's stock are you watching more closely as a signal for the AI sector overall? Hit reply with your take.

That's your intelligence briefing! 💪

Clayton

📧 Forward to your AI-curious friends

🔗 Connect: claytonstrategy.com