AI Trends — 12 June 2026
Your daily briefing on the most important developments in artificial intelligence.
🏦 The AI IPO Race Is On
The AI industry's biggest companies are rushing to go public, and the competition is fierce.
Anthropic confidentially filed its S-1 with the SEC on June 1, 2026, targeting a listing near its $965 billion post-money valuation — the result of closing a record $65 billion Series H round. The company is reporting a $47 billion annualized revenue run rate and is approaching its first profitable quarter. Goldman Sachs, Morgan Stanley, and JPMorgan are leading the process.
OpenAI filed its own confidential S-1 on June 8, openly announcing the move the same day: "We recently submitted a confidential S-1. We expect it to leak so we're just announcing it." Valued at $852 billion, OpenAI reported over $20 billion in ARR for 2025 — though internal projections show a $14 billion loss in 2026, with profitability not expected until 2029. A September–November 2026 listing window has been widely cited.
Investment bankers are advising both companies that there is a first-mover advantage to listing before the other. The race has begun.
🚀 SpaceX Goes Public: Largest IPO in History
SpaceX priced its IPO at $135 per share on June 11, raising approximately $75 billion at a $1.77 trillion valuation — surpassing Saudi Aramco's 2019 listing as the largest IPO in recorded history. Trading begins today on Nasdaq under the ticker SPCX.
The company absorbed xAI in an all-stock deal in February 2026, and Starlink alone generated $11.4B in revenue with $4.4B in segment operating income in 2025. Elon Musk will not sell a single share, retaining full voting control via a dual-class structure. Retail investors received an unusually large 30% allocation through Robinhood, Fidelity, and Charles Schwab.
🤖 Claude Opus 4.8: Smarter, More Honest, Faster
Anthropic launched Claude Opus 4.8 on May 28, 2026, with meaningful improvements in honesty and agentic performance:
- 4x less likely to let code flaws pass unremarked
- More likely to flag uncertainty; less prone to unsupported claims
- Fast mode runs at 2.5x the speed and is 3x cheaper than previous Opus fast mode
- Accompanied by Dynamic Workflows (research preview) for Claude Code — enabling hundreds of parallel subagents within a single session
Opus 4.8 is available at the same price as 4.7, and fast mode is accessible on Enterprise, Team, and Max plans. Anthropic also signaled that Mythos-class models will reach all customers "in the coming weeks."
📈 ChatGPT Hits 1 Billion Monthly Active Users
ChatGPT became the fastest application in history to reach 1 billion global monthly active users, crossing the milestone in May 2026 — roughly 3.5 years after its November 2022 launch. It surpassed the growth trajectories of Google Maps, TikTok, Instagram, and YouTube.
But the competitive picture is more nuanced. Claude's year-over-year MAU growth of approximately 640% dramatically outpaces ChatGPT's 62%, and US ChatGPT users who installed Claude in Q1 2026 spent 5% less time on ChatGPT one month later. One billion users is a floor, not a ceiling — the usage war is still being fought.
💰 The AI Price War Goes Structural
DeepSeek made its 75% price cut on V4 Pro permanent on May 23, 2026. New pricing: $0.003625–$0.87 per million tokens — well below OpenAI GPT-5 ($2.50/$10), Anthropic Claude Opus 4.7 ($5/$25), and Gemini 3.5 Flash ($0.15/$0.60). Running on Huawei Ascend 950 chips, DeepSeek avoids Nvidia supply constraints and US export controls entirely.
The price war is reshaping the mid-market. Enterprise contracts still go to Claude and GPT-5 on capability and compliance grounds. But startups, agencies, and independent developers — increasingly price-sensitive — are routing workloads to whatever is cheapest. DeepSeek is winning that segment aggressively.
☁️ OpenAI Integrates with Oracle Cloud
OpenAI officially announced on June 11 that enterprise customers can now access its frontier models and Codex through existing Oracle Universal Credits (UCM). Companies with multi-year Oracle cloud commitments — often worth tens to hundreds of millions — can deploy OpenAI models without a new vendor relationship.
This is a bigger deal than it appears: procurement friction, not model quality, is the real barrier to enterprise AI adoption. Removing it is a structural distribution win.
🔨 Microsoft Build 2026: Independence from OpenAI
Microsoft used Build 2026 (June 2) to announce seven in-house MAI models trained without OpenAI involvement — the first full exercise of rights restored after partnership restrictions were lifted in April 2026. Highlights:
- MAI-Thinking-1: reasoning flagship
- MAI-Code-1-Flash: 60% fewer tokens than comparable models, now live in GitHub Copilot
- MAI-Image-2.5: ranked above Gemini on Arena leaderboard
- MAI-Transcribe-1.5: SOTA transcription across 43 languages, 5x faster than rivals
The announcements coincided with the single largest government contract in Microsoft history — an AI deal covering 2.1 million US military service members and 770,000 civilian staff.
⚡ GitHub Copilot's Billing Change Sparks Backlash
GitHub Copilot switched from flat-rate plans to token-based billing on June 1, 2026. A single agentic coding session now consumes $30–40 in credits — 3–4x a Pro subscriber's entire $10 monthly allotment. Developer reaction on Reddit and Hacker News was swift and negative.
Microsoft's counter: flat-rate was unsustainable, with heavy users being subsidized by lighter ones. MAI-Code-1-Flash's 60% token efficiency directly addresses the credit burn problem. The all-you-can-eat era of AI coding assistance for individual developers is over.
🔭 What to Watch
- Gemini 3.5 Pro: Google promised it by end of June. Prediction markets have clustered around late June. When it ships, benchmarks will tell us how competitive H2 2026 really is.
- Anthropic vs. OpenAI IPO timing: Bankers say first-to-list wins the narrative. The weeks ahead will determine which company blinks first.
- DeepSeek's next move: A permanent price floor this low changes the economics for every AI-dependent startup. Expect more labs to respond.
Sources: Build Fast with AI · Anthropic · OpenAI · GitHub Blog · Reuters · Fortune